Arc fault protection method based on circuit breaker intelligent algorithm

Abstract

The invention relates to the technical field of electrical safety, and particularly discloses an arc fault protection method based on a circuit breaker intelligent algorithm, which comprises the following steps: S1, collecting a line current signal in real time, eliminating harmonic interference of a frequency converter from an original signal through an anti-aliasing filter, and converting the original signal into a digital signal; s2, processing the current signal by adopting a three-stage cascade wavelet analysis module, and outputting a dynamic peak-to-average ratio to S4; s3, detecting current disturbance based on a double-slide-rail window, outputting a disturbance positioning mark and a time-frequency feature to S4, judging a disturbance event, and positioning a starting point; s4, receiving the dynamic peak-to-average ratio of the S2 and the disturbance positioning mark and the time-frequency characteristics of the S3, performing path processing according to a characteristic analysis result, and adopting a direct interruption method and an electromagnetic radiation-current frequency domain cross validation method; s5, constructing an electric arc energy accumulation prediction model; according to the invention, the problems of weak arc leak detection, electromagnetic interference maloperation and delay are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical fault detection, and particularly relates to an arc fault protection method based on an intelligent algorithm of a circuit breaker. BACKGROUND

[0002] Among many power distribution safety hazards, arc fault, like a sword of Damocles, can release enough heat energy to ignite surrounding materials in an instant, leading to catastrophic electrical fires, causing immeasurable casualties and property losses. In the face of such a serious threat, traditional protection mechanisms such as thermal magnetic tripping or electronic overcurrent protection are not up to the task of arc recognition. The existing technology mainly relies on current amplitude threshold detection or simple waveform analysis, but these methods have essential limitations. When there are intermittent poor contact, aged wire insulation damage or carbonized path in the line, low-energy series arcs may be generated, such as <5A arcs caused by line aging and poor contact. The amplitude of these arc currents may not exceed the threshold set by the circuit breaker, and the waveform distortion is not enough to be effectively captured by conventional Fourier analysis. The traditional protection device may fail, allowing dangerous arcs to persist in the system. Common arc fault circuit breakers highly depend on the detection of high-frequency noise components in the current or the degree of harmonic distortion at a specific frequency. Such algorithms can be effectively verified in a laboratory pure load environment, but real working conditions are much more complex. Non-linear loads such as variable frequency drives, switching power supplies, energy-saving lamps, and electric motors generate a large amount of high-frequency harmonics and electromagnetic interference in use and operation states. The spectral characteristics of these signals often overlap and interfere with the signal components generated by early or weak fault arcs. This phenomenon directly leads to frequent false actions of the circuit breaker protection system, causing unnecessary interruptions of power supply and causing production and life disturbances. On the contrary, the more dangerous situation is that when there is a real arc risk, the system fails to accurately identify the weak and real arc characteristic signals in the complex interference background, leading to missed failures. Major safety hazards are thus created.

[0003] On the other hand, a deep exploration of the current situation of the industry can find that the existing arc fault protection technology generally has a deep bottleneck: lack of effective monitoring and trend prediction of the development process of fault energy. The arc fault detection module in the circuit breaker usually analyzes the signal characteristics in a single time segment in isolation and passively waits for certain preset characteristic quantities to cross the fixed threshold before reacting. However, the reality shows that the real dangerous arc that has the potential to cause a fire does not have all the destructive energy at the moment of explosion. On the contrary, such faults often originate from extremely weak or intermittent initial discharges. If the protection system cannot accurately identify the existence and development trend of these early low-energy discharges, it cannot capture the subtle abnormal trend changes in discharge intensity, frequency and duration. By the time the final accumulated energy triggers an electrical fire, the protection action is already too late. This lag and inability to recognize the qualitative change from the quantitative change in the development process of fault energy makes it difficult for the existing protection mechanism to move forward to the essential safety checkpoint. It needs to be emphasized that there are a large number of normal electromagnetic transient phenomena in complex circuits, and their frequency spectrum and time domain characteristics are highly similar to the initial stage low-energy arc discharge. For example, the transient overvoltage spark generated when inductive loads are switched, the commutation spark of old motor carbon brushes under certain working conditions, and even the inherent power switch oscillation in the working process of some high-frequency switching power supplies. These phenomena will produce noise characteristics or distorted waveforms similar to real series low-energy arcs. The existing technology based on a single or a small number of fixed frequency band energy distribution differences often faces a dilemma in such a complex and highly similar feature mixing scenario, and cannot reliably determine the existence of early low-energy arcs through the noise fog. It is even more difficult to scientifically quantify the potential harm and energy accumulation rate of these small discharge events in a specific line environment. When the fault arc is still in its low-energy incubation period, the existing technical solutions have strong limitations. The most critical point is that the arc at this stage has not yet formed a significant temperature rise to trigger a fire, but each weak discharge is silently accelerating the deterioration of the line, such as increasing the carbonization degree of the insulation material or damaging the surface state of the contact. This cumulative deterioration, like water dripping on stone, will eventually form a low-impedance channel, triggering a catastrophic arc explosion. Traditional protection methods have neither the awareness nor the ability to monitor this process and do nothing at its inception. This lack of monitoring and technical response to the cumulative trend of intermittent discharges is a key gap that needs to be filled in the field of arc fault protection. In summary, the serious difficulties encountered by traditional arc fault protection technology, especially the deep bottleneck of early low-energy arc detection and development trend prediction in complex load environments, have become a core barrier to the substantial improvement of electrical safety. If this bottleneck cannot be broken, the goal of improving the reliability of arc fault identification and preventing early electrical fires will be nothing more than an empty talk.

[0004] Therefore, the existing technology urgently needs an arc fault protection method based on an intelligent algorithm of a circuit breaker. SUMMARY

[0005] The purpose of the present application is to provide an arc fault protection method based on intelligent algorithm of circuit breaker, to solve the problem that the traditional arc protection device in the prior art cannot detect the micro-arc generated by the old line or poor contact when identifying early and weak potential arc fault, cannot identify the electromagnetic noise generated by normal equipment similar to the early arc signal, and lacks predictability for the micro-discharge that is accumulating damage.

[0006] To solve the above technical problems, the present application specifically provides the following technical solutions: An arc fault protection method based on intelligent algorithm of circuit breaker, comprising: S1, real-time acquisition of high-frequency current signal and dynamic noise reduction processing: real-time acquisition of line current signal, original signal elimination of anti-aliasing filter to eliminate frequency converter harmonic interference, converted into digital signal by MCU built-in ADC; S2, peak-to-average ratio calculation three-level wavelet decomposition: adopting three-level cascade wavelet analysis module to process current signal, outputting dynamic peak-to-average ratio to S4; S3, improved cumulative sum CUSUM algorithm disturbance positioning and time-frequency domain feature extraction: based on double sliding rail window to detect current disturbance, outputting disturbance positioning mark and time-frequency feature to S4, wherein the time-frequency domain feature includes current standard deviation σ and kurtosis β; S4, multi-modal hierarchical decision and protection execution: receiving dynamic peak-to-average ratio of S2 and disturbance positioning mark and time-frequency feature of S3, processing according to feature analysis result; S5, arc energy accumulation prediction: constructing arc energy accumulation prediction model, outputting processing result; wherein S2 and S3 are executed in parallel, and the output results are input into S4 synchronously, realized by dual-core MCU or real-time operating system.

[0007] Further, S1 comprises: real-time acquisition of line current signal by high-precision Rogowski coil current sensor, real-time calculation of load current change rate dI / dt, unit A / ms, when dI / dt>5 A / ms, active dynamic noise reduction algorithm is activated, the active dynamic noise reduction algorithm comprises: adopting adaptive wavelet threshold noise reduction, suppressing frequency converter harmonic interference by real-time adjustment of filter parameters, and improving signal-to-noise ratio to above 35 dB.

[0008] Further, S2 comprises: the wavelet analysis module uses db4 wavelet base function, the high-pass group coefficient is [-0.2304, 0.7148, -0.6309], the low-pass group coefficient is [0.0273, -0.0322, 0.1884], and the output is set to 50-100 kHz high-frequency component H1 as the first level, 25-50 kHz frequency band component H2 as the second level, and 12.5-25 kHz frequency band component H3 as the third level.

[0009] Further, S2 further comprises: calculating a dynamic peak-to-average ratio (PAR) in real time, the formula is: ; Wherein Take the maximum amplitude in H1-H3, is the three-channel average value, and the dynamic threshold is set to 15 dB + 0.3 x |dI / dt|, and the coefficient is adjusted to 0.5 when the load current standard deviation σ>5 A.

[0010] Further, the improved cumulative sum (CUSUM) algorithm in S3 uses a double sliding window, the front window W1 has a window length of 20 ms corresponding to 4000 points, and the rear window W2 has a window length of 10 ms corresponding to 2000 points, the current effective value mean μ of the W1 window is calculated, the offset threshold K=1.5 x μ is set, and the offset accumulation value in the W2 window is iteratively calculated, the formula is If >H and H=5 x K, it is determined that a disturbance event occurs, and the starting point is located.

[0011] Further, the locating of the starting point in S3 comprises: intercepting 20 ms of data before and after the disturbance point, calculating the current standard deviation σ and kurtosis β, taking the maximum point of the standard deviation as the center, if the kurtosis is the minimum value of the neighborhood 5 points, then it is marked as the starting point, the threshold σ>0.01, and the threshold β>3.5.

[0012] Further, the path processing according to the feature analysis result in S4 comprises: if the real-time dynamic peak-to-average ratio (PAR) is greater than 15 dB and the current standard deviation σ is greater than 0.01, a direct interruption method is adopted; if the real-time dynamic peak-to-average ratio (PAR) is in the range of [10 dB, 15 dB] and the kurtosis β is greater than 3.5, an electromagnetic radiation-current frequency domain cross verification method is started.

[0013] Further, the direct interruption method comprises: SA41, using a half-cycle event counter to accumulate, the half-cycle is 10 ms; SA42, if the accumulated events within 100 ms are greater than or equal to 8 times, a magnetic latching relay is triggered to trip; SA43, driving a MOSFET switch disconnection circuit, the conduction time is less than or equal to 10 ns; SA44, recording the fault waveform, and lighting a red LED indicator.

[0014] Further, the electromagnetic radiation-current frequency domain cross verification method comprises: SB41, synchronously collecting 200-300 MHz electromagnetic radiation signals, using a fourth-order Hilbert antenna, the gain is greater than or equal to 5dBi, and the sensitivity is -110 dBm; SB42, extract 10-30 kHz current energy E_c and 200-300 MHz electromagnetic energy E_m, construct feature vector [E_c, E_m]; SB43, calculate membership by pre-training FCM clustering model, fault clustering center is [0.85, 1.2], normal center is [0.15, 0.3]; SB44, if membership>0.8 and consecutive 3 windows meet the condition, trigger tripping, report fault location.

[0015] Further, the arc energy accumulation prediction model in S5 uses the formula; time interval T=10ms, if>200J / 100ms, determine as energy accumulation type fault, upload compressed fault feature data to cloud through wireless communication module with delay≤10ms, and output complete waveform to local monitoring terminal through wired communication interface, the wireless communication module is Bluetooth (BLE) protocol, the wired communication interface is RS485 industrial bus, transmission delay<10ms, to update FCM clustering center.

[0016] Further, it also includes deploying four-order Hilbert antenna to capture 200-300 MHz electromagnetic radiation signal, and using infrared thermal imaging sensor to monitor contact temperature rise, to construct current-electromagnetic-temperature three-modal sensing system.

[0017] Further, S5 also includes joint training of LS-SVM and dynamic threshold mechanism, setting arc type, load scene uniform distribution to form 8,000 groups of samples as training set, injecting 30% noise simulation electromagnetic interference into 2,000 groups of samples as test set, action time meets metallic short circuit≤30ms and series arc≤100ms in simulation environment and photovoltaic direct current system test, and positioning error in 200m cable is<10cm.

[0018] Compared with the prior art, the present application has the following beneficial effects: first, in view of the defect that the traditional detection method is slow in response to weak series arc, the extraction accuracy of high-frequency arc characteristics is improved by more than 40% based on the synergistic mechanism of three-level wavelet dynamic decomposition and kurtosis analysis. When the load current rate of change dI / dt>5 A / ms, the adaptive noise reduction algorithm adjusts the wavelet threshold in real time, and the signal-to-noise ratio under the harmonic interference of the frequency converter is improved to more than 35dB, so that the missed detection rate is compressed to less than 5% in the industrial frequency conversion equipment intensive scene; secondly, in view of the problem of misoperation caused by complex electromagnetic environment, the electromagnetic-current frequency domain cross verification path is innovatively designed: the 200-300MHz electromagnetic radiation signal and the 10-30kHz current energy are captured synchronously to construct the feature vector [E_c, E_m] input pre-training FCM clustering model. The model quantifies the membership difference between the fault clustering center [0.85, 1.2] and the normal center [0.15, 0.3], and still maintains 95% accuracy under 30dB noise background, reducing the false alarm rate by 60% compared with the traditional single current detection. By clarifying the data flow from S2 / S3 to S4, the problem of aggregation delay of parallel processing results is solved: S4 can directly call the PAR feature of S2 and the time domain positioning result of S3, avoiding repeated calculation; the decision response time is shortened to ≤100ms, which is improved by 19% compared with the scheme without clear flow direction; the double-channel data synchronization mechanism eliminates the conflict of inter-core communication, and the misoperation rate is reduced to 2.1%. More importantly, the present application breaks through the technical bottleneck of fault energy accumulation prediction. The improved cumulative and CUSUM algorithm adopts double sliding window to accurately locate the disturbance starting point, and combines the arc energy integral model, so that the system can issue a warning 500ms before the fault energy reaches the critical value. At the same time, the dynamic threshold mechanism updates the coefficient every 24 hours through Gaussian process regression, solving the adaptability problem of light / heavy load working conditions. The test shows that this scheme can compress the series arc action time to ≤100ms in the photovoltaic direct current system, the metallic short circuit response is ≤30ms, and the positioning error of 200m cable is <10cm, providing core technical support for accurate fault troubleshooting. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the following embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only exemplary, and those skilled in the art can also obtain other implementation drawings according to the provided drawings without creating any inventive labor.

[0020] Figure 1 The flowchart of the method according to the first embodiment of the present application. DETAILED DESCRIPTION

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The concepts involved in this application will first be described with reference to the accompanying drawings. It should be noted that the following descriptions of various concepts are only for the purpose of making the content of this application easier to understand and do not constitute a limitation on the scope of protection of this application; furthermore, the embodiments and features in the embodiments of this application can be combined with each other unless otherwise specified. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] Example 1 like Figure 1 As shown, this invention provides an arc fault protection method based on a circuit breaker intelligent algorithm, comprising the following steps: S1. Acquisition of high-frequency current signals and dynamic noise reduction processing: Real-time acquisition of line current signals. The original signal is filtered by an anti-aliasing filter to eliminate harmonic interference from the frequency converter, and then converted into a digital signal by the MCU's built-in ADC.

[0024] Specifically, the line current signal is acquired in real time by a high-precision Rogowski coil current sensor with a fixed sampling rate of 200 kHz and a sampling bit width of 16 bits. The original signal is filtered by an anti-aliasing filter (cutoff frequency 150 kHz, attenuation slope -60 dB / dec) to eliminate inverter harmonic interference, and then converted into a digital signal by the built-in ADC of the 32-bit MCU.

[0025] In S1, the sampled data is grouped in 20 ms intervals (corresponding to a complete power frequency cycle of a 50 Hz system), with each group containing 4000 sampling points. The load current change rate dI / dt is calculated in real time, with the unit being A / ms. When dI / dt > 5A / ms, the dynamic noise reduction algorithm is activated.

[0026] S2. Peak-to-Average Ratio (PAR) Calculation: Three-level wavelet decomposition and adaptive peak: A three-level cascaded wavelet analysis module (wavelet basis function db4, high-pass coefficient group: [-0.2304, 0.7148, -0.6309], low-pass coefficient group: [0.0273, -0.0322, 0.1884]) is used to process the current signal; S2 outputs the dynamic PAR to the S4 decision module as a criterion for high-frequency arc characteristics; PAR refers to the logarithmic ratio of the maximum amplitude of the high-frequency component to the root mean square value.

[0027] S3, improved cumulative sum (CUSUM) algorithm disturbance positioning and time-frequency domain feature extraction: based on double sliding window detection of current disturbance, output disturbance positioning marker and time-frequency features (standard deviation σ and kurtosis β) to S4 as event trigger basis; set the pre-window W1 window length to 20 ms (4000 points), the post-window W2 window length to 10 ms (2000 points), calculate the W1 window current effective value mean μ, set the offset threshold K = 1.5 μ, and iteratively calculate the offset accumulation value S in the W2 window n , the formula is , if >H (H = 5 K), it is determined as a disturbance event, and the starting point is positioned; wherein is the offset accumulation value of the current window, dimensionless; S n-1 is the offset accumulation value of the previous window, dimensionless; x n is the current sampling point current value, unit: ampere; μ is the W1 window (window length 20 ms) current effective value mean, unit: ampere; K is the offset threshold, unit: ampere, max(0, S n-1 +(x n - μ - K)) is the maximum value function, which ensures that the accumulation value is non-negative, dimensionless, and prevents negative offset from affecting detection.

[0028] Wherein S2 and S3 are executed in parallel, and the specific implementation mode is: the main controller starts S2 and S3 processing threads synchronously after completing S1 signal acquisition; under the dual-core MCU architecture, S2 and S3 are allocated to Core0 and Core1 respectively; single-core system realizes pseudo-parallel through RTOS, and the task switching period is ≤100 μs.

[0029] Specifically, an STM32H743 dual-core microcontroller is used: Core0 runs the S2 task: db4 wavelet transform is performed using the CMSIS-DSP library; Core1 runs the S3 task: improved CUSUM algorithm is implemented; shared memory area: 0x24000000-0x24000FFF; synchronization mechanism: data consistency is ensured through HSEM semaphore.

[0030] S4, multi-modal hierarchical decision and protection execution: receives the dynamic peak-to-average ratio of S2 and the disturbance positioning marker and time-frequency features of S3, and processes according to the feature analysis results, if the real-time calculated dynamic peak-to-average ratio PAR > 15 dB and the current standard deviation σ > 0.01 (regular electric arc), the direct interruption method is adopted; if the real-time calculated dynamic peak-to-average ratio PAR ∈ [10 dB, 15 dB] and the kurtosis β > 3.5 (weak electric arc), the electromagnetic radiation-current frequency domain cross verification method is adopted, which ensures that the parallel calculation results are efficiently converged to the decision layer, and reduces the instruction cycle delay (measured ≤0.8 ms).

[0031] S5, arc energy accumulation pre-judgment: build an arc energy accumulation prediction model, the formula used is: if ; wherein T is the time interval, the unit is second, fixed at 10 ms (0.01 s), each group of integral window length, if >200 J / 100ms, it is determined as energy accumulation type fault, through the BLE module to extract the compressed feature vector (including , positioning mark, clustering membership), upload to the cloud with ≤10ms delay; through the RS485 interface to output 20ms complete current waveform at 1Mbps baud rate, update FCM clustering center; wherein is the arc energy accumulation value, the unit is joule; summation operation, k from 1 to 8, represents the accumulation of 8 time intervals, covering 80ms window (T=10ms); integral operation, time from (k-1)T to kT, calculate the energy integral in each time interval; is the current square function, the unit is ampere², k is the time window index, dimensionless, the value range is 1-8, corresponding to the total length of 80ms.

[0032] The dual-channel redundant design includes: ① BLE transmission key features (data volume ≤256 bytes), ensure the real-time of cloud early warning; ② RS485 transmission original waveform (8KB data), used for local depth analysis; Specifically, the communication module implementation includes unlimited channel: using Nordic nRF5340 BLE 5.2 chip, configuring 1M PHY mode, data packet interval ≤7.5ms, transmitting feature vector[ , β, clustering membership]; wired channel: using ADI ADM2587E isolated RS485 transceiver, baud rate 1Mbps, transmitting 20ms / 4000 points original waveform (time-consuming 8.2ms).

[0033] The db4 wavelet base function uses Daubechies wavelet standard coefficient array, and the high-pass coefficient [-0.2304, 0.7148, -0.6309] conforms to the IEEE 1241-2010 specification.

[0034] Specifically, the direct interruption method includes the following steps: SA41, half-cycle event counter accumulates, half-cycle is 10 ms; SA42, if the cumulative events in 100 ms ≥8 times, trigger the magnetic latching relay to trip; SA43, drive MOSFET switch (on time ≤10 ns) to break the circuit; SA44, record the fault waveform and light the red LED indicator.

[0035] In particular, the electromagnetic radiation-current frequency domain cross-validation method comprises the following steps: SB41, synchronously collecting 200-300 MHz electromagnetic radiation signals using a fourth-order Hilbert antenna (gain > 5dBi, sensitivity -110 dBm); SB42, extracting 10-30 kHz current energy E_c and 200-300 MHz electromagnetic energy E_m to construct a feature vector [E_c, E_m]; SB43, calculating the membership degree through a pre-trained FCM clustering model (fault clustering center [0.85, 1.2], normal center [0.15, 0.3]); SB44, if the membership degree > 0.8 and the condition is met for 3 consecutive windows, triggering a trip and reporting the fault location.

[0036] The principle and effect of the above technical solution are as follows: the three-level wavelet dynamic peak average ratio detection is used to solve the problem of weak arc detection, the improved cumulative sum CUSUM algorithm is combined to position in time and frequency and resist electromagnetic noise interference, and the early energy accumulation prediction is realized by using electromagnetic-current feature fusion, which is 100 ms earlier than the traditional method; the output ends of the S2 wavelet decomposition module and the S3 disturbance positioning module are connected to the input end of the S4 decision module to form a "feature extraction-decision" straight link. The PAR value output by S2 is directly input to the threshold comparator of S4, and the disturbance marker output by S3 triggers the window interception unit of S4 to realize real-time fusion of dual-path data.

[0037] In one embodiment, the active dynamic noise reduction algorithm includes adaptive wavelet threshold noise reduction, real-time adjustment of filtering parameters to suppress inverter harmonic interference, improvement of signal-to-noise ratio to more than 35 dB, deployment of a fourth-order Hilbert antenna to capture 200-300 MHz electromagnetic radiation signals, use of an infrared thermal imaging sensor to monitor contact temperature rise, and construction of a current-electromagnetic-temperature three-modal sensing system.

[0038] In one embodiment, the positioning starting point includes intercepting 20 ms of data before and after the disturbance point, calculating the current standard deviation σ (threshold σ > 0.01) and kurtosis β (threshold β > 3.5), taking the maximum point of the standard deviation as the center, and if the kurtosis is the minimum value of the neighborhood of 5 points, marking it as the starting point.

[0039] In one embodiment, the current signal processing includes setting the output 50-100 kHz high-frequency component H1 as the first level, the output 25-50 kHz frequency band component H2 as the second level, and the output 12.5-25 kHz frequency band component H3 as the third level to calculate the dynamic peak average ratio (PAR) in real time; the formula used for the output 12.5-25 kHz frequency band component H3 to calculate the dynamic peak average ratio (PAR) in real time is: wherein Take the maximum amplitude in H1-H3, unit: ampere; H1-H3 is the root mean square average of the three channel amplitudes, unit: ampere.

[0040] The dynamic threshold is set to 15 dB + 0.3x|dI / dt|, the coefficient is adjusted to 0.5 when the load current standard deviation σ>5 A, the threshold coefficient is updated every 24 hours through the Gaussian process regression model, the formula is new threshold = original threshold x (1 +0.1x|actual state-predicted state|). When the load current standard deviation σ>5 A, the dynamic threshold coefficient switches to 0.5, which is fed back to S2 by S3 module in real time.

[0041] In one embodiment, S5 includes: S501, jointly train with LS-SVM and dynamic threshold mechanism, set arc type, load scene uniform distribution of 8,000 groups of samples as training set, inject 30% noise into 2,000 groups of samples to simulate electromagnetic interference as test set; S502, in the simulation environment (Matlab / Simulink) and photovoltaic DC system test, set the action time of metallic short circuit ≤30 ms and ≤ series arc action time ≤100 ms, the photovoltaic DC system test meets the positioning error <10 cm in 200 m cable.

[0042] The technical principle of the above scheme is: the Rogowski coil sensor can capture high-frequency current transient signals without distortion based on the principle of electromagnetic induction, the anti-aliasing filter is designed for the harmonic interference characteristics of nonlinear loads such as frequency converters, which blocks noise higher than the Nyquist frequency and prevents spectral aliasing. The dynamic noise reduction trigger mechanism is derived from the current mutation characteristics of fault arcs; db4 wavelet basis function has compact support and regularity, which is suitable for extracting high-frequency transient characteristics of arcs, dynamic peak-to-average ratio PAR quantifies the amplitude fluctuation of high-frequency components, fault arcs show amplitude mutation, and Gaussian process regression model dynamically adjusts threshold coefficient according to historical data to adapt to load fluctuation scene; double sliding window design is based on power frequency cycle characteristics, the front window establishes a stable baseline μ, and the rear window detects offset accumulation The kurtosis β>3.5 identifies the sharp peak characteristics of the current waveform; the standard deviation σ>0.01 reflects the discrete degree of the current, and assists in distinguishing the arc from the motor transient process; the direct interruption path is aimed at the typical parallel arc, the electromagnetic-current cross verification uses the electromagnetic radiation characteristics of the fault arc, and forms a redundant criterion with the current frequency domain characteristics, reduces the misoperation rate, the FCM clustering model quantifies the fault probability through membership calculation, and avoids misjudgment of a single sensor; the energy integral model reflects the joule heat accumulation of the fault duration, the threshold is set based on the ignition point experiment of the insulating material, the LS-SVM training sample covers 8 kinds of load scenes, such as capacitive / inductive load, motor, LED drive, ensures the model generalization ability, and the Bluetooth+RS485 dual-channel transmission meets the real-time early warning demand.

[0043] The technical effects of the above scheme are: the signal-to-noise ratio is improved to ≥35dB, the false alarm rate is significantly reduced, the adaptive noise reduction algorithm can suppress the wideband interference of frequency converters, switching power supplies and other equipment in industrial scenes, and the signal purity is improved; the detection sensitivity of weak series arcs is improved by 40%, in a 30dB noise environment, the Tektronix MDA800 is used to measure the false detection rate of 4.8%, and the dynamic threshold mechanism avoids the misoperation problem of the fixed threshold in light / heavy load working conditions; the positioning error is <0.1ms, which is 3 times higher than the accuracy of the traditional zero-crossing detection method, and the anti-electromagnetic noise interference ability is enhanced, and in a 30dB noise environment, it still maintains 95% accuracy; the typical arc action time is ≤100ms, the electromagnetic radiation auxiliary verification makes the weak arc recognition accuracy rate improve to 92%; the early warning of energy accumulation type fault is 500ms, which reduces the fire risk; in the photovoltaic direct current system test, the Fluke 1738 records an error of 8.5cm, which supports accurate fault troubleshooting.

[0044] In summary, through the closed-loop design of high-frequency signal processing (S1), multi-scale feature extraction (S2-S3), hierarchical decision (S4) and energy pre-judgment (S5), the problems of missed detection, misoperation and delay of traditional AFDD are solved, the weak arc detection rate is improved by using three-level wavelet decomposition and kurtosis analysis, the false alarm rate is reduced by electromagnetic radiation cross verification and FCM clustering, and the action time is compressed to within 100ms by CUSUM positioning and direct interruption path.

[0045] The principles and implementation manners of the present application are described herein by using specific examples, and the above example descriptions are only used to help understand the method of the present application and its core idea. The above descriptions are only preferred embodiments of the present application, and it should be pointed out that, due to the limited nature of the language expression, there are objectively infinite specific structures, and for ordinary skilled persons in the technical field, some improvements, refinements or changes can be made without departing from the principles of the present application, and the above technical features can also be combined in an appropriate manner; these improvements, refinements, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, shall be regarded as the protection scope of the present application.

Claims

1. An arc fault protection method based on a circuit breaker intelligent algorithm, characterized in that, include: S1. Real-time acquisition of high-frequency current signals and dynamic noise reduction: Real-time acquisition of line current signals, the original signal is filtered by an anti-aliasing filter to eliminate inverter harmonic interference, and then converted into a digital signal by the MCU's built-in ADC. S2. Peak-to-average power ratio calculation using three-level wavelet decomposition: The current signal is processed using a three-level cascaded wavelet analysis module, and the dynamic peak-to-average power ratio is output to S4. S3. Improved Cumulative Sum CUSUM Algorithm for Disturbance Localization and Time-Frequency Feature Extraction: Based on dual sliding window detection of current disturbance, the disturbance localization marker and time-frequency features are output to S4. The time-frequency features include the current standard deviation σ and kurtosis β. S4. Multimodal hierarchical decision-making and protection execution: Receive the dynamic peak-to-average power ratio of S2 and the disturbance location markers and time-frequency characteristics of S3, and process them according to the feature analysis results; S5. Arc energy accumulation prediction: Construct an arc energy accumulation prediction model and output the processing results; S2 and S3 are executed in parallel.

2. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 1, characterized in that, S1 includes: acquiring line current signals in real time using a high-precision Rogowski coil current sensor, calculating the load current change rate dI / dt in real time (in A / ms), and activating a dynamic noise reduction algorithm when dI / dt > 5 A / ms. The activation of the dynamic noise reduction algorithm includes using adaptive wavelet threshold noise reduction, suppressing inverter harmonic interference by adjusting filter parameters in real time, and improving the signal-to-noise ratio to over 35 dB.

3. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 1, characterized in that, S2 includes: The wavelet analysis module uses the db4 wavelet basis function, with high-pass group coefficients of [-0.2304, 0.7148, -0.6309] and low-pass group coefficients of [0.0273, -0.0322, 0.1884]. The output is set to 50–100 kHz, with high-frequency component H1 as the first level, 25–50 kHz frequency band component H2 as the second level, and 12.5–25 kHz frequency band component H3 as the third level.

4. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 3, characterized in that, S2 also includes: real-time calculation of dynamic peak-to-average power ratio (PAR), the formula is: ； in Take the largest amplitude value among H1–H3. The value is the average of the three channels, and the dynamic threshold is set to 15 dB + 0.3 × |dI / dt|. When the standard deviation of the load current σ > 5 A, the coefficient is adjusted to 0.

5.

5. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 1, characterized in that, The improved CUSUM algorithm in S3 uses a dual sliding window. The first window W1 has a window length of 20 ms and corresponds to 4000 points, while the second window W2 has a window length of 10 ms and corresponds to 2000 points. The mean effective value μ of the current in window W1 is calculated, and an offset threshold K = 1.5 × μ is set. The offset accumulation value within window W2 is then iteratively calculated. The formula is ,like If H = 5 × K, then it is determined to be a disturbance event, and the starting point is located.

6. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 5, characterized in that, The starting point for positioning in S3 includes extracting data 20 ms before and after the disturbance point, calculating the current standard deviation σ and kurtosis β, and taking the point with the largest standard deviation as the center. If its kurtosis is the minimum value of the 5 neighboring points, then it is marked as the starting point.

7. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 1, characterized in that, The path-based processing in S4 based on the feature analysis results includes: if the real-time calculated dynamic peak-to-average power ratio (PAR) is > 15 dB and the current standard deviation (σ) is > 0.01, then the direct interruption method is adopted; if the real-time calculated dynamic peak-to-average power ratio (PAR) is ∈ [10 dB, 15 dB] and the kurtosis (β) is > 3.5, then the electromagnetic radiation-current frequency domain cross-validation method is initiated.

8. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 7, characterized in that, The direct interruption method includes: SA41 uses a half-cycle event counter for accumulation, with a half-cycle of 10 ms; SA42: If the cumulative number of events is ≥8 within 100 ms, the magnetic latching relay will be triggered to trip. SA43, drives the MOSFET switching circuit, with a conduction time ≤10 ns; SA44 records the fault waveform and illuminates the red LED indicator.

9. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 7, characterized in that, The electromagnetic radiation-current frequency domain cross-verification method includes: SB41, synchronously acquires electromagnetic radiation signals in the 200–300 MHz range, using a fourth-order Hilbert antenna with a gain ≥5 dBi and a sensitivity of -110 dBm; SB42 extracts the current energy E_c from 10–30 kHz and the electromagnetic energy E_m from 200–300 MHz, and constructs the feature vector [E_c, E_m]. SB43 calculates membership degrees using a pre-trained FCM clustering model, with fault cluster centers at [0.85, 1.2] and normal cluster centers at [0.15, 0.3]. SB44: If the membership degree is greater than 0.8 and the condition is met for three consecutive windows, the trip is triggered and the fault location is reported.

10. The arc fault protection method based on circuit breaker intelligent algorithm according to claim 1, characterized in that, The formula used in the arc energy accumulation prediction model in S5 is: ;Time interval T=10ms, if >200 J / 100ms indicates an energy accumulation fault. Compressed fault characteristic data is uploaded to the cloud via a wireless communication module, and a complete waveform is output to the local monitoring terminal via a wired communication interface. The wireless communication module uses the Bluetooth Low Energy protocol, and the wired communication interface uses an RS485 industrial bus with a transmission delay of <10 ms, in order to update the FCM clustering center.

Images