Intelligent electro-optical disconnect system

Abstract

The application discloses an intelligent photoelectric breaking system and relates to the technical field of photoelectric breaking.The application realizes synchronous sampling and dynamic characteristic extraction of optical signals and current signals through a time base generation and characteristic extraction module, adaptively generates an asymmetric time correlation window in combination with a time window construction module, and performs hierarchical marking triggering and state migration according to the arc light signal intensity, the current amplitude change rate and the time correlation window by a multi-criterion fusion and state machine module, so that pure light interference and real arc fault can be effectively distinguished.Meanwhile, a protection execution and fault record module realizes pre-tripping, formal tripping, reignition detection and hierarchical fault storage, and a channel failure protection reconstruction module automatically switches to a pure overcurrent or pure arc light delay protection mode when a sensor or a mutual inductor fails, so that the adaptive capability, the protection reliability and the system fault tolerance of arc fault identification are remarkably improved under complex working conditions and transient changes.

Description

Technical Field

[0001] This invention relates to the field of photoelectric switching technology, specifically to an intelligent photoelectric switching system. Background Technology

[0002] In actual arc fault protection scenarios for distribution cabinets, the arc fault is dynamically affected by multiple factors, and existing technologies have the following shortcomings: First, existing methods do not quantify and couple the dynamic changes in the time difference between the arrival time of the arc pulse and the candidate time of the current change with fault distance, CT transient transmission delay, and zero-crossing nonlinear distortion. This leads to the fixed window failing to detect faults when they are located at the end of the line or near the zero-crossing point due to excessive time difference, and misjudging when there is a slight time shift due to normal load fluctuations, resulting in a serious lack of adaptability of the judgment criteria. Second, existing solutions lack a dynamic baseline update mechanism that is linked to real-time operating conditions. They cannot effectively distinguish between faults and normal disturbances during large current transient processes such as motor startup or transformer inrush current. They are also prone to failure to trip due to fixed threshold failure after the arc sensor ages or the fiber optic cable attenuates. Furthermore, they lack adaptive weighting and state machine hierarchical transition logic based on the physical characteristics of arc plasma, resulting in the inability to accurately distinguish between pure optical interference and real faults. There is no dynamic upgrade and feedback correction path between alarms and trips, leading to a long-term mismatch between protection reliability and speed.

[0003] Therefore, there is an urgent need for an intelligent photoelectric switching system that can dynamically construct an asymmetric time correlation window based on the feature vector of the real-time arc rise steepness and the derivative of the current zero crossing point, learn the dynamic baseline and fluctuation range based on historical data, realize multi-criteria thresholdless adaptive fusion and state machine hierarchical transfer, and compensate for CT nonlinear distortion and channel delay online, so as to solve the above problems. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent photoelectric interruption system, which solves the problem that existing intelligent photoelectric interruption methods cannot adapt to different working conditions and changes in the transient characteristics of electric arcs due to fixed-time correlation windows and static criteria.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent photoelectric switching system, comprising: The time base generation and feature extraction module is used to synchronously sample the arc light sensor and the current transformer to obtain a first digital code stream and a second digital code stream. Arc light features are extracted based on the first digital code stream, and current features are extracted based on the second digital code stream.

[0006] The time window construction module is used to construct time-related windows based on arc light characteristics and current characteristics.

[0007] The multi-criteria fusion and state machine module is used to mark and trigger the process based on arc light characteristics and current characteristics, combined with arc light signal intensity, current amplitude change rate and time correlation window. The state machine transitions between monitoring mode, arc light alarm mode and tripping mode according to the triggered mark.

[0008] The protection execution and fault recording module is used to output pre-trip signals and formal trip signals in trip mode, detect reignition and accumulate reignition counts. The fault process is stored hierarchically and associated.

[0009] The channel failure protection reconfiguration module is used to switch to pure overcurrent protection mode or pure arc delay protection mode respectively when the arc sensor channel or current transformer channel failure is detected.

[0010] The present invention has the following beneficial effects: This invention achieves synchronous sampling and dynamic feature extraction of optical and current signals through a time base generation and feature extraction module. Combined with a time window construction module, it adaptively generates an asymmetric time correlation window. The multi-criteria fusion and state machine module performs hierarchical marking triggering and state transition based on the arc signal intensity, current amplitude change rate, and time correlation window, effectively distinguishing between pure optical interference and real arc faults, avoiding misjudgments and omissions caused by fixed windows and static criteria. At the same time, the protection execution and fault recording module realizes pre-trip, formal trip, reignition detection, and hierarchical fault storage. The channel failure protection reconstruction module automatically switches to pure overcurrent or pure arc delay protection mode when the sensor or transformer fails. Thus, it significantly improves the adaptive capability of arc fault identification, protection reliability, and system fault tolerance under complex operating conditions and transient changes.

[0011] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0012] Figure 1 This is a flowchart of the intelligent photoelectric switching system of the present invention.

[0013] Figure 2 This is a flowchart illustrating the process of obtaining the first and second digital code streams in the intelligent photoelectric switching system of the present invention.

[0014] Figure 3 This is a flowchart illustrating the hierarchical storage and association of fault processes in the intelligent photoelectric switching system of the present invention. Detailed Implementation

[0015] Please see Figure 1 This invention provides a technical solution: an intelligent photoelectric switching system, comprising: The time base generation and feature extraction module is used to synchronously sample the arc light sensor and the current transformer to obtain a first digital code stream and a second digital code stream. Arc light features are extracted based on the first digital code stream, and current features are extracted based on the second digital code stream.

[0016] Given the lack of a synchronization reference between the arc signal and the current signal, the complex electromagnetic environment inside the distribution cabinet, the different signal transmission path lengths of the arc sensor and the current transformer, and the different response times of the signal conditioning circuits, even using the same clock source, the actual time when the signal arrives at the processing unit will fluctuate due to path differences. Furthermore, the entire process from the generation of a fault arc to its detection is extremely short; any relative time error caused by asynchronous sampling could directly mask the physical causal relationship between the actual arc-leading current or the current-leading arc. Therefore, a unified and traceable time reference must be established at the very beginning.

[0017] Among them, such as Figure 2 As shown, the process of obtaining the first and second digital bitstreams is as follows: After the system powers on, the microcontroller first configures the operating frequency of the global time base generator. This operating frequency is typically more than twice the highest frequency of the sampled signal to satisfy the Nyquist sampling theorem. Once configured, the global time base generator outputs a continuous square wave clock signal. This clock signal is connected to the external clock input pins of the photoelectric conversion unit and the signal acquisition unit via transmission lines. To ensure that both units receive the clock edge simultaneously, identical drive buffers are inserted in both paths, and serpentine routing is used to compensate for differences in line length. During transmission, the global time base generator also generates a synchronization reset pulse to reset the internal sampling counters of both units to zero, thus ensuring perfect alignment of the first sampling moment.

[0018] An arc sensor receives light radiation from within the distribution cabinet and transmits the optical signal to a photoelectric conversion unit via optical fiber. Inside the photoelectric conversion unit, a photodiode generates a weak photocurrent under illumination. This photocurrent is converted into an initial analog voltage by a transimpedance amplifier. Since this initial analog voltage may be lower than the effective input range limit of the analog-to-digital converter (ADC), it is fed into a programmable gain amplifier. The programmable gain amplifier automatically selects an appropriate amplification factor based on the gain coefficient pre-stored during system self-test or the peak voltage feedback in real time, ensuring that the output analog voltage fills as much of the ADC's full-scale range as possible without saturation. The adjusted analog voltage is then fed into a first ADC, which samples, holds, and quantizes the analog voltage on each rising edge of the synchronous sampling clock, outputting a first digital code stream.

[0019] The secondary side of the current transformer outputs a small current signal proportional to the primary current. This small current signal generates a corresponding analog voltage through a precision sampling resistor inside the signal acquisition unit. Since the fault current can span two orders of magnitude from normal load to short-circuit current, the amplitude range of the analog voltage is very large. Therefore, it also needs to be adaptively adjusted by a programmable gain amplifier. The programmable gain amplifier and the transimpedance amplifier use the same model and configuration to ensure gain consistency between the two channels. The adjusted analog voltage is sent to the second analog-to-digital converter (ADC). The second ADC shares the same synchronous sampling clock signal with the first ADC, performing analog-to-digital conversion at the same sampling rate and resolution to obtain the second digital code stream.

[0020] The global time base generator generates a global time tag at each sampling time and appends the global time tag to the corresponding first digital bitstream and second digital bitstream respectively.

[0021] The first digital stream, the second digital stream, and their respective global timestamps are stored in a shared ring buffer, which stores data from at least the most recent two power frequency cycles.

[0022] The global time base generator refers to an independent clock chip used to generate the reference time pulse for the entire system. It is obtained by outputting a square wave signal of a fixed frequency through a hardware crystal oscillator or phase-locked loop. The value (such as 10MHz or 20MHz) determines the limit of the sampling time resolution. The higher the value, the smaller the time interval between adjacent sampling points, and the better it can capture the arc pulse leading edge at the microsecond level.

[0023] A programmable gain amplifier is an operational amplifier circuit whose amplification factor can be adjusted by software. Different gains are obtained by changing the feedback resistor value through a configuration register. The value of the gain factor (such as 1, 2, 4, 8 times) is used to adjust the analog voltage signal to the optimal input range of the analog-to-digital converter (such as 0-5V). If the gain is too high, it will easily saturate; if it is too low, the signal-to-noise ratio will be insufficient.

[0024] Each value in the first digital bitstream represents the instantaneous arc intensity at a sampling moment. It is obtained by continuously reading the output register of the analog-to-digital converter, and its value range (e.g., 0-4095) corresponds to the minimum to maximum light intensity; a larger value indicates a stronger arc at that moment. Each value in the second digital bitstream represents the instantaneous current value at a sampling moment. The sign of the value indicates the current direction, and the absolute value is proportional to the primary current.

[0025] A shared circular buffer is a circular memory queue used to simultaneously store a first digital bitstream, a second digital bitstream, and their global timestamps. Data is written to a fixed-size buffer in chronological order through a memory management unit, and the oldest data is overwritten when the buffer is full.

[0026] The power frequency period refers to the fundamental period of the AC current or voltage in a power system, typically 20ms (50Hz system) or 16.67ms (60Hz system). It is calculated by detecting the zero-crossing interval in the second digital code stream, and its magnitude (20ms or 16.67ms) determines the reference duration of all subsequent dynamic baselines and time windows.

[0027] This embodiment uses a global time base generator to simultaneously send synchronous sampling clock signals to the photoelectric conversion unit and the signal acquisition unit, and adds a global time tag to each sampling point, eliminating the sampling time offset between the optical channel and the current channel. Simultaneously, the time-tagged dual-channel digital code stream is stored in a shared ring buffer and retained for at least two power frequency cycles, ensuring that the historical waveform before the arc occurs can be traced back, thus allowing the system to retrieve the signal state of the power frequency cycle preceding the fault after the fault occurs.

[0028] It should be noted that when an arc fault occurs, the arc light signal has an extremely fast rising edge, while the interference light generated by normal lighting or switching operations rises relatively slowly. If the exact moment of arc onset cannot be accurately captured, any subsequent time-related criteria will lose their physical basis. Furthermore, while the ambient light inside the distribution cabinet may be close in intensity to a weak arc, ambient light typically does not exhibit a steep rising edge and simultaneously correlate with the power frequency cycle phase in a very short time. Therefore, it is necessary to simultaneously utilize the instantaneous change trend and cycle phase comparison to distinguish between the real arc light and the background light. Specifically, the process of extracting arc light features based on the first digital code stream is as follows: The first digital bitstream is read from the shared ring buffer of the synchronous sampling and time base generation module.

[0029] For each sampling point in the first digital bitstream, the value of the sampling point is calculated as the arc intensity.

[0030] Calculate the first difference between the arc intensity of the sampling point and the arc intensity of the previous sampling point in the same power frequency cycle, and the second difference between the arc intensity of the sampling point and the arc intensity at the same phase position in the previous power frequency cycle.

[0031] When three consecutive sampling points meet the set conditions, the global time tag of the first sampling point among the three consecutive sampling points is marked as the arrival time of the arc pulse. The set conditions are: the first difference of the three consecutive sampling points is greater than zero, the second difference of the three consecutive sampling points is greater than zero, and the arc intensity of the three consecutive sampling points shows a monotonically increasing trend.

[0032] Centered on the sampling point corresponding to the arrival time of the arc pulse, the arc signal intensity values ​​and their respective global time stamps of seven sampling points—the three sampling points before the sampling point, the sampling point itself, and the three sampling points after the sampling point—are extracted from the shared circular buffer. These seven data points are considered as coordinate points on a plane, with the global time stamp on the x-axis and the arc signal intensity on the y-axis. A univariate linear regression is performed on these seven points to calculate the slope of the regression line. The slope is calculated as follows: the sum of the products of the x and y coordinates of all sampling points minus the product of the sum of the x and y coordinates divided by the number of sampling points, then divided by the sum of the squares of the x coordinates minus the square of the sum of the x coordinates divided by the number of sampling points. The calculated slope is the arc rise steepness.

[0033] The arrival time of the arc pulse and the steepness of the arc rising edge are used as arc characteristics.

[0034] The first difference can be positive or negative. A positive value indicates that the intensity at the current point is stronger than the previous point; a negative value indicates a weakening. Three consecutive first differences greater than zero indicate that the arc light intensity is continuously increasing. A positive second difference indicates that the light intensity of that phase in the current cycle is higher than the same phase in the previous cycle, helping to eliminate interference from periodic background light. Three consecutive second differences greater than zero indicate that the arc light intensity is continuously increasing relative to the background light in a normal cycle.

[0035] The magnitude of the arrival time of the arc pulse indicates the absolute time from the start of the rising edge of the arc.

[0036] The magnitude of the arc rise steepness value indicates how fast the arc intensity rises per unit time. The larger the value, the faster the arc forms and the more intense the energy release. The unit is light intensity per microsecond.

[0037] This embodiment calculates the first and second differences at each sampling point and jointly determines that the first and second differences of three consecutive sampling points are all greater than zero, and the arc signal intensity monotonically increases. This allows for the effective differentiation of the rapid rising edge of the real arc from background light fluctuations, gradual changes in ambient light, or periodic interference without relying on any preset thresholds. This avoids mislabeling of the arc pulse arrival time due to non-faulty light sources. Simultaneously, it uses linear regression to calculate the arc rising edge steepness using seven sampling points before and after the arc pulse arrival time, quantifying the transient rate of the arc formation process into a continuously changing and physically meaningful characteristic parameter. This allows the system to directly determine the steepness of the arc signal without relying on fixed slope thresholds or empirical values.

[0038] Considering that the current waveform exhibits significant changes distinct from normal loads during fault current occurrences, but these changes take many forms, a single feature cannot cover all fault types. Furthermore, current transformers may saturate at the initial stage of a fault current, leading to distortion of the secondary current waveform. In this case, traditional amplitude or RMS value criteria become ineffective, necessitating reliance on the waveform's morphological characteristics near the zero-crossing point for judgment. Additionally, the unique zero-crossing extinction and reignition phenomena of arc faults cause nonlinear distortion of the current near the zero-crossing point of each half-wave. The degree of this distortion is closely related to the fault type and can serve as an important basis for distinguishing between faults and normal loads. In summary, the process of extracting current features based on the second digital code stream is as follows: Used to read the second digital bitstream from the shared ring buffer of the synchronous sampling and timing generation module. For each sample point in the second digital bitstream, perform the following calculation: The rate of change of current amplitude is obtained by dividing the difference between the instantaneous current value at the current sampling point and the instantaneous current value at the previous sampling point by the sampling interval.

[0039] The difference between the instantaneous current value at the current sampling point and the instantaneous current value at the same phase position of the previous power frequency cycle is used as the instantaneous overcurrent characteristic.

[0040] The points where the second digital bitstream value changes from negative to positive or from positive to negative are marked as zero-crossing points.

[0041] The extreme points of the first derivative of the second digital code stream are detected, and the derivative value corresponding to the first point after the current zero crossing point whose absolute value of the derivative is greater than twice the absolute value of the maximum derivative within a preset number of sampling points centered on the current zero crossing point is taken as the current mutation amount. At the same time, the corresponding global time tag is taken as the candidate time of the current mutation.

[0042] The time offset of the current sampling point from the nearest zero-crossing point is taken as the power frequency phase.

[0043] Two sampling points are taken before and after each zero-crossing point, and the corresponding four current amplitude change rate values ​​are extracted and arranged in time order to obtain the zero-crossing derivative feature vector.

[0044] The current abrupt change and its corresponding candidate time of the current abrupt change, power frequency phase, and zero-crossing derivative eigenvector are used as current features.

[0045] The magnitude of the rate of change of current amplitude can be positive or negative; a positive value indicates an increase in current, and a negative value indicates a decrease in current. The larger the absolute value, the more drastic the current change. In the initial stage of a fault current, this value will be much greater than the rate of change of normal load.

[0046] The magnitude of the instantaneous overcurrent characteristic value reflects the increment of current relative to the normal cycle; a positive value indicates an increase in current, and a negative value indicates a decrease. A consistently large positive value indicates a persistent overcurrent in the circuit.

[0047] The global time stamp of the zero-crossing point is used to determine the power frequency phase. The zero-crossing point occurs at the end of each half-wave, and the interval is equal to half a power frequency cycle.

[0048] The first derivative is the rate of change of the instantaneous value of current over time. It is calculated in the same way as the rate of change of current amplitude, i.e., by dividing the difference between adjacent sampling points by the sampling interval, obtained through point-by-point calculation. The magnitude of the value indicates the rate of change of current, and the sign indicates the direction of change. Near the zero-crossing point of the current, the derivative of normal load current is small and symmetrical, while the derivative of arc fault current will show abnormal spikes or asymmetry.

[0049] The absolute value of the derivative refers to the absolute value of the first derivative. Its magnitude represents the amplitude of the rate of change of current, regardless of direction. It is used to compare the intensity of the derivative near different zero crossings to determine whether there are abnormal abrupt changes.

[0050] The current surge is the derivative value at the point where the absolute value of the first derivative after a zero-crossing is greater than twice the absolute value of the derivative near the previous zero-crossing. The magnitude of the value indicates the severity of the current surge; a larger value indicates a steeper rise in fault current.

[0051] Candidate times for current surges are obtained by recording the global time stamp of the sampling point. The value represents the absolute time of the current surge, which is used for subsequent timing matching.

[0052] Power frequency phase refers to the time offset of the current sampling point from the nearest zero-crossing point, expressed in units of time (e.g., microseconds) or angles. The numerical range is from 0 to half a power frequency cycle (e.g., 0 to 10 ms in a 50Hz system). The value indicates the current position within the half-wave of the power frequency; the phase is smaller near the zero-crossing point, and the phase at the peak is close to half the cycle.

[0053] The zero-crossing derivative eigenvector reflects the rate of change of current near the zero-crossing point. Under normal load, these four values ​​are symmetrical and relatively small; under arc fault conditions, they may exhibit asymmetry or larger values.

[0054] This embodiment extracts current amplitude change rate, instantaneous overcurrent characteristics, zero-crossing point, current surge and candidate time of current surge, power frequency phase, and zero-crossing derivative feature vector from the second digital code stream, providing multi-dimensional current-side features for arc fault determination without relying on any preset thresholds. Specifically, the current amplitude change rate can capture instantaneous abrupt changes in current, suitable for rapid short-circuit faults; the instantaneous overcurrent characteristics reflect the overall current rise, suitable for slowly developing overcurrent faults; the current surge and its candidate time are obtained through derivative comparison, avoiding false triggering caused by load fluctuations; and the power frequency phase and zero-crossing derivative feature vector are specifically used to describe the nonlinear distortion behavior of AC arcs near the zero-crossing point, a key feature for distinguishing arc faults from linear load inrush currents.

[0055] The time window construction module is used to construct time-related windows based on arc light characteristics and current characteristics.

[0056] Because there is an inherent physical timing relationship between the arc light signal and the current signal generated by an arc fault, this timing relationship is not fixed but is affected by multiple factors such as fault distance, loop inductance, current transformer transmission characteristics, and arc reignition phase. If a fixed time window is used to determine the correlation between arc light and current, then when the fault occurs near the zero crossing point or is far from the sensor, the fixed window will either miss the real fault or misjudge normal disturbances.

[0057] It should also be noted that current transformers exhibit small-signal nonlinear transmission errors near the zero-crossing point. At this point, the reliability of the current characteristics decreases, necessitating dynamic adjustment of the tolerance for current lead time based on the degree of distortion. The rise steepness of the arc directly reflects the energy release rate of the arc; the faster it forms, the more likely the optical signal is to significantly lead the current signal. Therefore, the asymmetry of the time window must adaptively change with the rise steepness.

[0058] Taking the above considerations into account, the specific process for constructing a time-related window is as follows: Based on the output power frequency phase, determine whether the arrival time of the arc pulse is located in the region near the zero crossing point. The region near the zero crossing point is defined as having a time offset from any zero crossing point that is less than an adaptive boundary. The adaptive boundary is equal to three times the reciprocal of the arc pulse rising edge steepness.

[0059] If the arrival time of the arc pulse is located near the zero-crossing point, the degree of nonlinear distortion of the second digital code stream near the zero-crossing point is calculated using the derivative eigenvector of the zero-crossing point. The degree of nonlinear distortion is defined as the ratio of the absolute value of the sum of the two current amplitude change rates after the zero-crossing point to the absolute value of the sum of the two current amplitude change rates before the zero-crossing point. Based on the degree of nonlinear distortion, the upper limit of the allowable light lead time and the upper limit of the allowable current lead time are adjusted. Specifically, the degree of nonlinear distortion is equal to the ratio of the sum of the two current amplitude change rates after the zero-crossing point to the sum of the two current amplitude change rates before the zero-crossing point. The degree of nonlinear distortion is used as a weighting factor to expand the upper limit of the allowable light lead time to (one plus the weighting factor) times the initial value, while reducing the upper limit of the allowable current lead time to (one minus half of the weighting factor) times the initial value. The initial value is determined by the sampling period and the response time of the programmable gain amplifier.

[0060] Based on the arc rise edge steepness and a proportionality coefficient obtained through self-learning, the upper limit of the allowable light lead time is adjusted proportionally, and the upper limit of the allowable current lead time is adjusted inversely: the upper limit of the allowable light lead time is directly proportional to the arc rise edge steepness, and the proportionality coefficient is the ratio of the average photoelectric delay to the average arc rise edge steepness obtained through system self-learning after the most recent fault clearing. The proportionality coefficient is updated during each fault clearing confirmation mode in the system self-learning process. The upper limit of the allowable current lead time is inversely proportional to the arc rise edge steepness, and the inverse proportionality coefficient is the same self-learning ratio.

[0061] Based on the time difference between the arrival time of the arc pulse and the candidate time of the current abrupt change, an exponentially weighted moving average correction is applied to the upper limits of the allowable light lead time and the allowable current lead time: The time difference between the arrival time of the arc pulse and the candidate time of the current abrupt change is calculated. If the time difference is positive and its absolute value is greater than the current allowable light lead time upper limit, the allowable light lead time upper limit is increased by one-tenth of the difference between the time difference and the allowable light lead time upper limit. If the time difference is negative and its absolute value is greater than the current allowable current lead time upper limit, the allowable current lead time upper limit is increased by one-tenth of the difference between the absolute value of the time difference and the allowable current lead time upper limit. The above correction is smoothed using an exponentially weighted moving average, with the smoothing coefficient being the ratio of the variance to the mean of the time differences over the most recent ten power frequency cycles.

[0062] When no tripping flag is triggered within a consecutive preset number (e.g., 20) power frequency cycles, and the average time difference is less than a certain proportion (e.g., 0.5 times) of the current allowable upper limit, the allowable light lead time upper limit and the allowable current lead time upper limit are gradually brought closer to the initial value by an exponentially weighted moving average.

[0063] The starting boundary is obtained by subtracting the upper limit of the allowable current lead time from the arrival time of the arc pulse, and the ending boundary is obtained by adding the upper limit of the allowable light lead time to the arrival time of the arc pulse. The starting boundary and the ending boundary constitute a time correlation window.

[0064] Among them, the degree of nonlinear distortion is close to negative one under an ideal sine wave (because the signs of the rates of change are opposite); when arc distortion occurs, the ratio will deviate from negative one, for example, due to reignition, the absolute value of the sum of the last two rates of change increases abnormally, and the absolute value of the ratio is greater than one.

[0065] The upper limit of the allowable light lead time refers to the maximum time that the arrival time of the arc pulse is allowed to precede the candidate time of the current change in the time correlation window, expressed in time. The numerical value represents the maximum time difference that the system can tolerate for the optical signal to arrive earlier than the current signal; the larger the value, the wider the window extends in the direction of light lead.

[0066] The upper limit of the allowable current lead time refers to the maximum time that the candidate moment of the current abrupt change is allowed to precede the arrival time of the arc pulse within the time correlation window, expressed in time. The numerical value represents the maximum time difference that the system can tolerate for the current signal to arrive earlier than the optical signal; the larger the value, the wider the window extends in the direction of current lead.

[0067] The initial values ​​refer to the base values ​​of the allowed upper limits of light lead time and current lead time without any compensation, adjustment, or correction, and are preset by the system. The values ​​are generally between a few microseconds and tens of microseconds, and the larger the value, the greater the inherent hardware delay.

[0068] The proportionality coefficient is dimensionless and reflects the statistical relationship between the photoelectric delay and the intensity of the arc in a specific distribution cabinet environment. A larger value indicates a longer light lead time at the same steepness. It should be noted that the system determines the fault is cleared after outputting a formal trip signal and the reignition count reaches zero. The microcontroller extracts the absolute value of the time difference between the arrival time of the arc pulse and the candidate time of the current abrupt change from the shared ring buffer, denoted as the photoelectric delay for this fault; it also extracts the arc rise steepness. The system maintains a delay queue and a steepness queue containing a maximum of ten recent faults, updated using a first-in, first-out (FIFO) principle. The updated average delay is equal to the sum of all values ​​in the delay queue divided by the queue length; the average steepness is similarly calculated, with the proportionality coefficient equal to the average delay divided by the average steepness. If the number of historical faults is less than ten, the factory default value is used as the proportionality coefficient.

[0069] The average optical-electrical delay refers to the average time difference between the arrival time of the arc pulse and the corresponding candidate time of current change during multiple faults counted in the system's self-learning process. It is obtained by summing the absolute values ​​of this time difference in each fault and dividing by the number of faults. The value is measured in time and represents the statistically average degree of optical signal lead.

[0070] The self-learning process is triggered after each protection action (i.e., after the formal trip signal output) and when the subsequent reignition counter is 0. The system first calculates the absolute value of the time difference between the arrival time of the arc pulse and the candidate time of the current surge based on data recorded in the shared ring buffer, using this as the current photo-electric delay T_delay. If multiple arc pulse arrival times or multiple candidate times of current surges exist during the current fault, the system selects the arrival time of the earliest arc pulse and the corresponding candidate time of the current surge that falls within the time association window constructed by that arc pulse, and calculates the absolute value of their time difference as the current photo-electric delay T_delay. Simultaneously, the arc rising edge steepness K_slope of the current fault is obtained. The system maintains a delay queue and a steepness queue containing at most the most recent N (e.g., 10) faults. After each fault, the arithmetic mean T_avg of the delay queue and the arithmetic mean K_avg of the steepness queue are calculated, and the updated proportional coefficient α = T_avg / K_avg. If the number of historical faults is less than N, the system-preset initial proportional coefficient is used. The exponentially weighted moving average (EMA) is a smoothing correction algorithm used to progressively adjust the upper limits of allowable light lead time and allowable current lead time. It is obtained by multiplying newly observed deviation values ​​by a weighting coefficient and then averaging them with older values. The smoothing coefficient is determined by the ratio of the variance to the mean of the time difference over the most recent ten power frequency cycles. A larger ratio indicates more drastic time difference fluctuations, a larger smoothing coefficient, and a faster system response; conversely, a smaller smoothing coefficient indicates a more gradual change.

[0071] The numerical range of the time-related window extends from the start boundary to the end boundary, and the total window length is the upper limit of the allowed light lead time plus the upper limit of the allowed current lead time.

[0072] This embodiment utilizes power frequency phase and adaptive boundaries to determine whether the arc is located near the zero-crossing point, and accordingly adjusts the upper limits of the allowable light lead time and the allowable current lead time to compensate for the small-signal nonlinear distortion of the current transformer near the zero-crossing point, avoiding timing misjudgments caused by CT transmission delay distortion. Furthermore, based on the arc rise steepness and self-learning coefficient, the upper limit of the allowable light lead time is adjusted proportionally and the upper limit of the allowable current lead time is adjusted inversely, ensuring that the window shape matches the physical characteristics of the arc in real time. This prevents the missed detection of rapid arcs due to an excessively narrow window, and avoids the misjudgment of normal load fluctuations due to an excessively wide window. Finally, the window boundary is corrected online using an exponentially weighted moving average, and the tolerance range is dynamically updated using the observed arc-current time difference. This allows the system to continuously approach the optimal window parameters during long-term operation, thereby simultaneously improving the reliability and speed of protection.

[0073] The multi-criteria fusion and state machine module is used to mark and trigger the process based on arc light characteristics and current characteristics, combined with arc light signal intensity, current amplitude change rate and time correlation window. The state machine transitions between monitoring mode, arc light alarm mode and tripping mode according to the triggered mark.

[0074] During normal operation of the distribution cabinet, both arc light and current signals exhibit inherent background fluctuations. The amplitude and frequency of these fluctuations vary depending on the operating conditions, making it impossible to distinguish between normal and abnormal conditions using a fixed threshold. Although the arc light intensity generated by an arc fault is much higher than the background light, the intensity increase varies depending on the type of fault. Using an absolute threshold may lead to missed detection of weak arcs or misjudgment of strong interference.

[0075] Furthermore, relying solely on instantaneous triggering of light intensity anomalies may lead to misjudgments due to noise spikes. Therefore, the anomaly needs to persist for a certain period before confirmation. However, this duration should not be fixed but rather correlated with the steepness of the arc's rising edge. The faster the arc forms, the shorter the confirmation time should be to ensure rapid response. The rate of change of current amplitude also exhibits normal fluctuations, the amplitude of which is much smaller than the rate of change of fault current. However, the normal fluctuation range varies greatly across different circuits, thus requiring a dynamic baseline rather than a fixed threshold. The coordination between the candidate moment of current abrupt change and the time correlation window is the core of timing criteria. However, directly using the rate of change of current amplitude may be affected by noise; therefore, the amount of current abrupt change is introduced as a more reliable basis for timing matching.

[0076] Specifically, the flag triggering process is as follows: The median of the arc light intensity of all sampling points within the past power frequency cycle, read from the shared ring buffer, is used as the dynamic arc light baseline. The relative arc light intensity is obtained based on the ratio of the current real-time arc light intensity to the dynamic arc light baseline.

[0077] An abnormal arc light signal intensity is determined when the relative arc light intensity exceeds twice the maximum relative arc light intensity fluctuation range in the last ten power frequency cycles. A pure arc light marker is triggered when the continuous duration of the abnormal arc light intensity is greater than or equal to the length of a dynamic time window, where the length of the dynamic time window is equal to twice the reciprocal of the current arc light rising edge steepness. Furthermore, the abnormal state must persist within a continuous time window, the length of which is dynamically equal to twice the reciprocal of the current arc light rising edge steepness. Within the dynamic time window, if the relative arc light intensity at each sampling point meets the above conditions, a pure arc light marker is triggered.

[0078] The median of the absolute values ​​of the current amplitude changes of all sampling points within the past power frequency cycle, which are read from the shared ring buffer, is used as the dynamic rate of change baseline. The relative rate of change is obtained based on the ratio of the absolute value of the current real-time current amplitude change rate to the dynamic rate of change baseline.

[0079] When the relative change rate is greater than three times the maximum relative change rate in the last ten power frequency cycles, it is determined to be an abnormal current amplitude change rate. When both the abnormal arc intensity and the abnormal current amplitude change rate are met, the arc and current change rate flag is triggered.

[0080] Arcing and current mutation timing markers are triggered when the candidate time of current mutation falls between the start and end boundaries of the time association window and the absolute value of the current mutation is greater than five times the baseline of the dynamic rate of change.

[0081] Among them, the value of the dynamic arc light baseline represents the normal background light level under the current working conditions. It is automatically updated as the ambient light changes slowly. The larger the value, the stronger the ambient light.

[0082] The numerical value of relative arc intensity is dimensionless. It is close to 1 under normal conditions, but much greater than 1 when an arc occurs. The larger the value, the greater the increase in arc intensity relative to the background.

[0083] The maximum relative arc light intensity fluctuation range over the last ten power frequency cycles refers to the maximum value among the ten differences calculated for each of the ten power frequency cycles (the difference between the maximum and minimum relative arc light intensity values ​​within that cycle). This value is obtained through cycle-by-cycle statistical analysis, and its magnitude represents the maximum possible fluctuation range of the relative arc light intensity under normal operating conditions. A larger value indicates more severe background light fluctuations.

[0084] The dynamic time window refers to the length of time used to check whether the abnormal arc signal intensity persists. Its length is dynamically equal to twice the reciprocal of the current arc rise edge steepness. It is obtained by calculating the reciprocal of the arc rise edge steepness and then multiplying it by two. The unit of measurement is time. The steeper the arc rise edge, the shorter the window length; the smaller the steepness, the longer the window. The length of the window directly affects the protection's speed of response.

[0085] The baseline value of the dynamic rate of change represents the typical amplitude of the current change rate under normal operating conditions. The larger the value, the more severe the normal load fluctuation of the circuit.

[0086] The relative rate of change is dimensionless, and is close to 1 under normal conditions, but is much greater than 1 under fault conditions.

[0087] The maximum relative rate of change fluctuation range in the most recent ten power frequency cycles refers to the relative rate of change fluctuation range calculated for each of the past ten power frequency cycles (the difference between the maximum and minimum relative rate of change in that cycle), and the maximum value among these ten differences is taken. It is obtained through statistical analysis on a cycle-by-cycle basis, and its magnitude represents the maximum possible fluctuation range of the relative rate of change under normal operating conditions.

[0088] This embodiment achieves adaptive anomaly detection for arc light signal intensity and current amplitude change rate without relying on any preset thresholds by using a dynamic arc light baseline, a dynamic rate of change baseline, and comparisons based on historical fluctuation ranges. Specifically, the pure arc light marker requires the anomaly to persist within a dynamic time window whose length is twice the reciprocal of the arc light's rising edge steepness. This allows the system to automatically adjust the confirmation time according to the speed of arc formation, avoiding speed loss or insufficient anti-interference capability caused by fixed delays. The arc light plus current change rate marker uses the simultaneous fulfillment of light intensity anomalies and rate of change anomalies as trigger conditions, effectively eliminating ambient light interference with only transient light intensity changes and no current changes. The arc light plus current abrupt change timing marker uses a time correlation window and current abrupt change magnitude for timing matching, improving reliability in environments with current waveform distortion or noise. The hierarchical design of these three markers enables the system to distinguish between different scenarios such as pure light interference and real faults, fundamentally solving the problems of misjudgment and missed judgment caused by fixed thresholds and single criteria.

[0089] Furthermore, considering that arcing events in the distribution cabinet may be of different types—some are merely momentary strong light interference without accompanying current abnormalities, while others are actual arcing faults requiring immediate isolation—if all arcing events directly trip the circuit breaker, it will lead to frequent false trips; if all events only trigger alarms, the fault cannot be isolated in a timely manner. Therefore, a tiered approach is necessary. Specifically, the process by which the state machine transitions between monitoring mode, arcing alarm mode, and tripping mode based on the triggered flag is as follows: The initial state is monitoring mode.

[0090] In monitoring mode, when the pure arc light marker is triggered and neither the arc light current change rate marker nor the arc light current sudden change timing marker is triggered, the state machine switches from monitoring mode to arc light alarm mode and outputs an arc light alarm signal.

[0091] In monitoring mode, when the arc current change rate marker is triggered or the arc current sudden change timing marker is triggered, the state machine directly switches from monitoring mode to trip mode.

[0092] In arc alarm mode, when the pure arc marker is continuously triggered and neither the arc current change rate marker nor the arc current sudden change timing marker is triggered, the state machine remains in arc alarm mode and repeatedly outputs the arc alarm signal.

[0093] In arc alarm mode, when the pure arc marker stops triggering, the state machine returns from arc alarm mode to monitoring mode.

[0094] In arc alarm mode, when the arc current change rate marker or the arc current sudden change timing marker is triggered, the state machine switches from arc alarm mode to trip mode.

[0095] This embodiment divides the system status into three levels: monitoring mode, arc alarm mode, and tripping mode, and defines clear transition rules, directly solving the problem that a single protection action cannot distinguish between arc events of different severity. The specific benefits are as follows: First, when a pure arc marker is triggered alone, it only switches to arc alarm mode and outputs an alarm signal, without directly tripping, avoiding false tripping caused by ambient light interference or non-faulty arcs, thus improving power supply reliability. Second, in arc alarm mode, continuous monitoring of current characteristics is maintained. Once the arc plus current change rate marker or the arc plus current change sequence marker is triggered, it immediately switches to tripping mode, ensuring that real faults are not missed, achieving a seamless connection between alarm and tripping. Third, when the pure arc marker stops triggering, it automatically returns to monitoring mode, avoiding the tediousness of manual reset and the accumulation of long-term alarm states. Fourth, the entry conditions for tripping mode cover two combination markers; regardless of which is met, rapid tripping is possible, improving adaptability to different fault types.

[0096] The protection execution and fault recording module is used to output pre-trip signals and formal trip signals in trip mode, detect reignition and accumulate reignition counts. The fault process is stored hierarchically and associated.

[0097] It should be noted that the operation of the circuit breaker's tripping coil requires a certain amount of time for the excitation current to build up. Applying the full tripping current directly may result in excessive coil surge or delayed operation. Therefore, a pre-trip signal needs to be output for pre-excitation to reduce the subsequent tripping time. Furthermore, the timing of the formal tripping signal output should not be immediate upon entering the tripping mode, as the fault current may not have fully formed or may be near its zero-crossing point. Premature tripping may prevent the arc from being reliably extinguished. It is necessary to wait for the fault current to cross zero and confirm its continued existence to ensure the circuit breaker breaks the circuit when the current naturally crosses zero, thus reducing arc energy.

[0098] It should also be noted that reignition may occur during the circuit breaker contact breaking process, meaning that the gap between the contacts is broken down and an electric arc is generated again. In this case, a trip signal needs to be repeatedly output to drive the circuit breaker to operate again. If reignition occurs more than twice, it indicates that the circuit breaker itself may be damaged, and further tripping will be ineffective and may even aggravate the fault. Therefore, it is necessary to lock out and trigger an alarm. Reignition detection should be based on the recurrence of arc pulses and current surges, rather than just the presence of current, to avoid misinterpreting normal current fluctuations as reignition.

[0099] Specifically: When the state machine enters the trip mode, a pre-trip signal is output, which is used to pre-excite the circuit breaker's tripping coil. Simultaneously with the pre-trip signal output, the second digital code stream is continuously read from the shared ring buffer, and the current zero-crossing points in the second digital code stream are detected. When two consecutive current zero-crossing points are detected, and the instantaneous current value in the second digital code stream is still greater than one-tenth of the average fault-free current in the most recent power frequency cycle, a formal trip signal is output. The formal trip signal continues to be output until the instantaneous current values ​​at three consecutive sampling points after detecting a current zero-crossing point in the second digital code stream are all less than the average fault-free current.

[0100] The process of detecting and accumulating the reignition count is as follows: During the output of the formal trip signal, if a new arc pulse arrival time is detected again, and simultaneously a new current surge candidate time falls within the time correlation window, it is determined that the circuit breaker contact has reignited. The formal trip signal is then repeatedly output, and the reignition count is accumulated. When the reignition count exceeds two, all subsequent trip outputs are blocked, and a circuit breaker failure alarm signal is output. The blocked state can only be cleared by an external manual reset signal or a system power failure restart. During the blocked state, all new trip requests are ignored, and the circuit breaker failure alarm signal is continuously output.

[0101] This embodiment waits for two consecutive current zero-crossing points and then checks if the instantaneous current is still greater than one-tenth of the average fault-free current. This avoids prematurely terminating the tripping due to misjudging the fault as cleared near the natural current zero-crossing point, ensuring that a formal tripping is only performed when the fault current actually exists. Simultaneously, during the formal tripping signal period, the timing correlation between the arrival time of new arc pulses and new current mutation candidate times is continuously monitored to determine circuit breaker contact re-ignition. This accurately identifies re-breakdown during the breaking process and intervenes by repeatedly outputting the formal tripping signal and accumulating the re-ignition count. When the re-ignition count exceeds two, subsequent tripping outputs are blocked, and a circuit breaker failure alarm signal is output, preventing equipment damage and system impact caused by invalid repeated tripping, and improving the self-diagnostic capability and safety of the protection system.

[0102] Because arc faults occur and are cleared in a continuous process, the data characteristics and diagnostic value differ at different stages. Storing all data indiscriminately would result in a massive data volume and make it difficult to locate key information during fault analysis. Therefore, if... Figure 3 As shown, the process of storing and associating fault processes in layers is as follows: The fault process is divided into three consecutive time layers: the pre-trigger layer, the fault evolution layer, and the post-fault layer.

[0103] The pre-trigger layer stores data from the start of the power frequency cycle before the arrival of the arc pulse to the end of the arrival of the arc pulse, including the first digital code stream, the second digital code stream, the arc signal intensity, the rate of change of current amplitude, the instantaneous overcurrent characteristics, the current mutation amount, and the global time tag within the corresponding time interval.

[0104] The fault evolution layer stores data from the arrival time of the arc pulse to the end of the formal trip signal output, including the first digital code stream, the second digital code stream, the arc signal intensity, the rate of change of current amplitude, the instantaneous overcurrent characteristics, the current mutation amount, the global time tag, the transition type and transition time of each state machine, the curves of the change of the start and end boundaries of the time association window over time, and the trigger status of the pure arc mark, the arc plus current change rate mark, and the arc plus current mutation timing mark for each sampling point.

[0105] The post-fault layer stores data from the end of the formal trip signal output until the first digital code stream recovers to the dynamic arc light baseline range and the second digital code stream recovers to the fault-free current average range, including all data within the corresponding time interval, as well as reignition event records and circuit breaker operation time.

[0106] By linking the data of the pre-triggering layer, fault evolution layer, and post-fault layer through a unique event association identifier, multiple reignitions or multiple arc pulses generated by the same physical fault can be attributed to the same event set.

[0107] Differential compression is used for storage, storing only the difference between the current sampling point value and the previous sampling point value, as well as the sampling point number corresponding to the difference.

[0108] This embodiment solves the problems of chaotic fault data recording and inability to trace the complete process by dividing the fault process into three consecutive time layers: a pre-trigger layer, a fault evolution layer, and a post-fault layer, and linking them with event association identifiers. The advantages are as follows: The pre-trigger layer independently stores the raw data of the previous power frequency cycle before the fault, providing a historical comparison benchmark for determining whether the arc initiation time is truly abnormal. The fault evolution layer not only stores waveform data but also synchronously stores each transition type and time of the state machine, the time association window boundary change curve, and the trigger status of each marker, allowing fault analysts to reproduce the entire execution process of the protection logic and verify whether the criteria were correctly triggered. The post-fault layer specifically records the residual process from the end of the trip signal to the system returning to normal, including reignition events and circuit breaker operating times, facilitating the evaluation of the circuit breaker's breaking performance. Through unique event association identifiers, multiple reignitions or multiple arc pulses generated by the same physical fault are grouped into the same event set, avoiding misanalysis caused by fragmented records. The differential compression method significantly reduces storage space requirements, making long-term, high-sampling-rate fault recording possible.

[0109] The channel failure protection reconfiguration module is used to switch to pure overcurrent protection mode or pure arc delay protection mode respectively when the arc sensor channel or current transformer channel failure is detected.

[0110] It should be noted that arc flash sensors and current transformers, as input devices of the protection system, may experience faults such as fiber optic aging, lens contamination, CT winding breakage, or magnetic core saturation after long-term operation. If the system still relies on these signals for criterion fusion when these channels fail, it will lead to protection failure or false tripping. Moreover, the degradation of arc flash sensor channels is usually manifested by significant inconsistencies in the dynamic baseline values ​​output by multiple sensors within the same protection area, while the background light of each sensor should be basically consistent under normal operating conditions. Therefore, the degraded channel can be identified by calculating the variance. In addition, when a current transformer channel fails, it cannot provide accurate current characteristics, but the current ratio of adjacent circuits during the same load period is similar. Abnormal channels can be found through lateral comparison. Therefore, when a certain type of channel completely fails, the protection system should not stop working, but should degrade to use another type of signal to continue providing basic protection, while recording the failure status word for maintenance and repair.

[0111] The process for detecting failure of the arc sensor channel or current transformer channel is as follows: At the end of each power frequency cycle, if the variance of the dynamic baseline value of the arc intensity corresponding to all arc sensors in the same protection area exceeds a set multiple (e.g., five times) of the average variance of the most recent set number (e.g., one hundred) power frequency cycles, it is determined that there is a risk of arc sensor channel failure in the protection area.

[0112] When all arc sensor channels within a certain protection area are deemed to be at risk of failure, the system automatically switches to pure overcurrent protection mode. In pure overcurrent protection mode, tripping judgment is made based on the rate of change of current amplitude and instantaneous overcurrent characteristics. When the absolute value of the rate of change of current amplitude exceeds ten times the dynamic rate of change baseline for three consecutive power frequency cycles, or when the instantaneous overcurrent characteristics exceed the dynamic overcurrent baseline for three consecutive power frequency cycles, a tripping signal is output and the arc channel complete failure status word is recorded at the same time. The dynamic overcurrent baseline is twice the peak current of the previous power frequency cycle.

[0113] Meanwhile, when the ratio of the second digital code stream corresponding to the current transformer in the same circuit to the second digital code stream corresponding to the current transformer in the adjacent circuit exceeds the set multiple (e.g., five times) of the average ratio of the most recently set number (e.g., one thousand) of power frequency cycles, the current transformer channel is determined to be abnormal.

[0114] When the current transformer channel of a protected circuit is determined to be faulty, it automatically switches to the pure arc delay protection mode. In the pure arc delay protection mode, the tripping judgment is based solely on the arc intensity, the arrival time of the arc pulse, and the pure arc marker in the marker triggering process. When the pure arc marker is continuously triggered within four consecutive power frequency cycles, a tripping signal is output and the current channel complete loss status word is recorded at the same time.

[0115] The protected area refers to several independent spaces within the distribution cabinet, divided by physical location. Each area is equipped with one or more arc flash sensors. The system configuration parameters define the area number, and the sensor data from different areas are independently compared for variance.

[0116] Adjacent circuits refer to feeder circuits that share the same busbar as the protected circuit and are electrically distanced within the same distribution cabinet or between adjacent distribution cabinets. Their current waveforms should exhibit similar trends under normal load. The system pre-sets the transformer channel numbers for adjacent circuits through configuration parameters.

[0117] When all arc sensor channels within the same protection area are deemed to be faulty, and the current transformer channels in the same protected circuit are also deemed to be faulty, the system switches to safe standby mode, outputs a serious system fault alarm, and executes protection actions according to preset strategies (such as activating backup protection or delay tripping).

[0118] This embodiment utilizes the comparison of the variance of the dynamic baseline values ​​of all arc flash sensors within the same protection area with the historical average variance to automatically identify the degradation of individual or all arc flash sensors without thresholding, avoiding erroneous judgments caused by sensor output distortion. When all arc flash sensor channels fail, it automatically switches to a pure overcurrent protection mode, using the rate of change of current amplitude and instantaneous overcurrent characteristics to continue tripping protection, ensuring fault clearing capability even when the optical signal is missing. Simultaneously, by statistically analyzing the historical ratio of currents in adjacent circuits, abnormalities in current transformer channels can be detected, and when one fails, it switches to a pure arc flash delay protection mode, using continuous triggering of the pure arc flash marker to achieve delayed tripping, avoiding complete loss of protection function due to current signal loss. Both degradation modes record status words, providing clear fault location information for subsequent maintenance.

[0119] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0124] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent photoelectric switching system, characterized in that, include: The time base generation and feature extraction module is used to synchronously sample the arc light sensor and the current transformer to obtain the first digital code stream and the second digital code stream. Arc light features are extracted based on the first digital code stream, and current features are extracted based on the second digital code stream. The time window construction module is used to construct time-related windows based on arc light characteristics and current characteristics; The multi-criteria fusion and state machine module is used to mark and trigger the process based on arc light characteristics and current characteristics, combined with arc light signal intensity, current amplitude change rate and time correlation window. The state machine transitions between monitoring mode, arc light alarm mode and tripping mode according to the triggered mark. The protection execution and fault recording module is used to output pre-trip signals and formal trip signals in trip mode, detect reignition and accumulate reignition counts; Store and associate fault processes in hierarchical layers; The channel failure protection reconfiguration module is used to switch to pure overcurrent protection mode or pure arc delay protection mode respectively when the arc sensor channel or current transformer channel failure is detected.

2. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of obtaining the first and second digital bitstreams is as follows: Simultaneously send a synchronous sampling clock signal to both the photoelectric conversion unit and the signal acquisition unit; The photoelectric conversion unit converts the optical signal from the arc sensor into an analog voltage signal, which is then adjusted by a programmable gain amplifier and converted into the first digital code stream. The signal acquisition unit converts the secondary current signal from the current transformer into an analog voltage signal, which is then adjusted by a programmable gain amplifier and converted into a second digital code stream. Global timestamps are appended to the corresponding first and second digital bitstreams, respectively.

3. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of extracting arc light features based on the first digital code stream is as follows: Calculate the first difference between the arc intensity of the sampling point and the arc intensity of the previous sampling point in the same power frequency cycle, and the second difference between the arc intensity of the sampling point and the arc intensity at the same phase position in the previous power frequency cycle; When the first difference and the second difference corresponding to three consecutive sampling points meet the set conditions, the global time tag of the first sampling point among the three consecutive sampling points is marked as the arrival time of the arc pulse. Obtain the arc intensity values ​​of three sampling points before and three sampling points after the arrival time of the arc pulse, and use the linear regression method to calculate the slope of the arc signal intensity as a function of the global time stamp, which is defined as the arc rising edge steepness. The arrival time of the arc pulse and the steepness of the arc rising edge are used as arc characteristics.

4. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of extracting current features based on the second digital code stream is as follows: The extreme points of the first derivative of the second digital code stream are detected, and the derivative value corresponding to the first point after the current zero crossing point whose absolute value of the derivative is greater than twice the absolute value of the maximum derivative within a preset number of sampling points centered on the current zero crossing point is taken as the current mutation amount, and the corresponding global time tag is taken as the candidate time of the current mutation. The time offset of the current sampling point from the nearest zero-crossing point is taken as the power frequency phase. Two sampling points are taken before and after each zero crossing point, and the corresponding four current amplitude change rate values ​​are extracted and arranged in time order to obtain the zero-crossing derivative feature vector. The current abrupt change and its corresponding candidate time of the current abrupt change, power frequency phase, and zero-crossing derivative eigenvector are used as current features.

5. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of constructing a time correlation window based on arc light characteristics and current characteristics is as follows: Based on the power frequency phase of the current characteristics, if it is determined that the arrival time of the arc pulse in the arc characteristics is located in the region near the zero crossing point, then the degree of nonlinear distortion is calculated using the zero-crossing derivative eigenvector in the current characteristics, and the upper limit of the allowable light lead time and the upper limit of the allowable current lead time are adjusted based on the degree of nonlinear distortion. Based on the arc rise steepness in the arc characteristics and the proportional coefficient obtained by self-learning, the upper limit of the allowed light lead time is adjusted proportionally and the upper limit of the allowed current lead time is adjusted inversely. Based on the time difference between the arrival time of the arc pulse and the candidate time of current abrupt change in the current characteristics, the upper limit of the allowable light lead time and the upper limit of the allowable current lead time are corrected by an exponential weighted moving average. The starting boundary is obtained by subtracting the upper limit of the allowable current lead time from the arrival time of the arc pulse, and the ending boundary is obtained by adding the upper limit of the allowable light lead time to the arrival time of the arc pulse. The starting boundary and the ending boundary constitute a time correlation window.

6. The intelligent photoelectric switching system according to claim 1, characterized in that, The flag triggering process is as follows: The median of the arc light intensity of all sampling points in the past power frequency cycle is used as the dynamic arc light baseline, and the ratio of the current real-time arc light intensity to the dynamic arc light baseline is used as the relative arc light intensity. When the relative arc intensity is greater than twice the maximum relative arc intensity fluctuation range in the last ten power frequency cycles, it is determined to be an abnormal arc signal intensity. When the continuous duration of the abnormal arc intensity is greater than or equal to the length of a dynamic time window, a pure arc mark is triggered, wherein the length of the dynamic time window is equal to twice the reciprocal of the current arc rising edge steepness. The median of the absolute values ​​of the current amplitude change rates of all sampling points in the past power frequency cycle is taken as the dynamic change rate baseline. The relative change rate is obtained based on the ratio of the absolute value of the current real-time current amplitude change rate to the dynamic change rate baseline. When the relative change rate is greater than three times the maximum relative change rate in the last ten power frequency cycles, it is determined to be an abnormal current amplitude change rate. When both the abnormal arc intensity and the abnormal current amplitude change rate are met, the arc and current change rate flag is triggered. Arcing and current mutation timing markers are triggered when the candidate time of current mutation falls between the start and end boundaries of the time association window and the absolute value of the current mutation is greater than five times the baseline of the dynamic rate of change.

7. The intelligent photoelectric switching system according to claim 6, characterized in that, The process by which the state machine transitions between monitoring mode, arc alarm mode, and trip mode based on the triggered flag is as follows: The initial state is monitoring mode; In monitoring mode, when the pure arc light marker is triggered and neither the arc light current change rate marker nor the arc light current sudden change timing marker is triggered, the state machine switches from monitoring mode to arc light alarm mode and outputs an arc light alarm signal. In monitoring mode, when the arc current change rate marker is triggered or the arc current sudden change timing marker is triggered, the state machine directly switches from monitoring mode to trip mode. In arc alarm mode, when the pure arc marker is continuously triggered and neither the arc current change rate marker nor the arc current sudden change timing marker is triggered, the state machine remains in arc alarm mode and repeatedly outputs the arc alarm signal. In arc alarm mode, when the pure arc marker stops triggering, the state machine returns from arc alarm mode to monitoring mode; In arc alarm mode, when the arc current change rate marker or the arc current sudden change timing marker is triggered, the state machine switches from arc alarm mode to trip mode.

8. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of detecting reignition and accumulating the reignition count is as follows: During the output of the formal trip signal, if a new arc pulse arrival time is detected again, and at the same time a new current change candidate time falls within the time correlation window, it is determined that the circuit breaker contact reignites, the formal trip signal is repeatedly output and the reignition count is accumulated; when the reignition count exceeds two, all subsequent trip outputs are blocked, and a circuit breaker failure alarm signal is output.

9. The intelligent photoelectric switching system according to claim 1, characterized in that, The process of storing and associating fault processes in layers is as follows: The fault process is divided into three consecutive time layers: the pre-trigger layer, the fault evolution layer, and the post-fault layer. The pre-trigger layer stores data from the start of the power frequency cycle before the arrival of the arc pulse to the end of the arrival of the arc pulse; The fault evolution layer stores data from the arrival time of the arc pulse to the end of the formal trip signal output; The post-fault layer stores data from the end of the formal trip signal output until the first digital code stream recovers to the range of the dynamic arc light baseline and the second digital code stream recovers to the range of the average fault-free current. By linking the data of the pre-triggering layer, fault evolution layer and post-fault layer through a unique event association identifier, multiple reignitions or multiple arc pulses generated by the same physical fault can be attributed to the same event set. Differential compression is used for storage, storing only the difference between the current sampling point value and the previous sampling point value, as well as the sampling point number corresponding to the difference.

10. The intelligent photoelectric switching system according to claim 1, characterized in that, The process for detecting failure in the arc sensor channel or current transformer channel is as follows: At the end of each power frequency cycle, if the variance of the dynamic baseline value of the arc intensity corresponding to all arc sensors in the same protection area exceeds the set multiple of the average variance of the most recent set number of power frequency cycles, it is determined that there is a risk of arc sensor channel failure in the protection area. When all arc sensor channels within a certain protection area are deemed to be at risk of failure, the system automatically switches to pure overcurrent protection mode. In pure overcurrent protection mode, tripping judgment is made based on the rate of change of current amplitude and instantaneous overcurrent characteristics. Meanwhile, when the ratio of the second digital code stream corresponding to the current transformer in the same circuit to the second digital code stream corresponding to the current transformer in the adjacent circuit exceeds the set multiple of the average ratio of the most recent set number of power frequency cycles in the same load time period, the current transformer channel is determined to be abnormal. When the current transformer channel of a protected circuit is determined to be faulty, it automatically switches to the pure arc delay protection mode. In the pure arc delay protection mode, the tripping judgment is based solely on the arc intensity, the arrival time of the arc pulse, and the pure arc marker in the marker triggering process.

Images