Intelligent monitoring method and system for a fuse

Abstract

The present application relates to the technical field of intelligent power distribution network monitoring, and more particularly to a fuse intelligent monitoring method and system, comprising: an edge computing module acquires a micro-disturbance tilt angle sequence of a fuse carrier under normal conditions through a three-axis tilt sensor, extracts a phase space feature to generate a mechanical stiffness index and outputs a fatigue warning; simultaneously, a transient voltage sequence of a secondary side energy storage capacitor of a CT power taking unit is collected for differential operation to obtain a voltage mutation factor; when the mutation factor reaches a threshold value, stop listening, and based on a thermal breakdown threshold value, an optimal limit flash time window is obtained; the switch is closed according to the time window, the surge energy is injected into the LED light source to irradiate the hot color plastic layer on the surface of the contact, and the red and blue reflected voltages are obtained; the mutation factor is introduced into a compensation model to correct the reflected voltage, the color transfer index is calculated and is latched to a non-volatile memory, and the CT secondary side is simultaneously short-circuited. The present application realizes power-down latching of pre-disaster extreme value data and complete hardware bypass self-protection.

Description

Technical Field

[0001] This invention relates to the field of intelligent power distribution network monitoring technology, and in particular to an intelligent monitoring method and system for fuses. Background Technology

[0002] As a core protection switch for medium-voltage power distribution networks, drop-out fuses are prone to abnormal contact resistance and overheating due to long-term environmental corrosion of their outdoor contacts. Currently, the industry commonly uses miniature current transformers for power monitoring. When a sudden severe short circuit occurs, generating a transient high-voltage surge, existing devices typically immediately activate a hardware bypass switch to physically short-circuit the secondary side of the transformer to prevent damage to low-voltage components. However, this conventional design has a fatal flaw: at the moment of the short circuit, the bypass short-circuit causes the main control circuit to momentarily lose power and stop working, resulting in the permanent loss of extreme thermal degradation data on the contact surface prior to the fault—a crucial piece of evidence. Specifically, if bypass is executed, the temperature measurement variables in the buffer stack are erased as the capacitor discharges; if the power supply is maintained and the data is latched, the high-voltage spike will vaporize the main control chip within microseconds. Existing technology has consistently failed to achieve a balance between protecting hardware during surges and preserving evidence of the fault. Summary of the Invention

[0003] To overcome the above shortcomings, the present invention provides a method and system for intelligent monitoring of fuses, which aims to improve the problem that most traditional fuse monitoring devices use direct bypass discharge, resulting in the permanent loss of extreme thermal evidence during fault outbreaks due to instantaneous power loss of the system.

[0004] In a first aspect, the present invention provides the following technical solution: a method for intelligent monitoring of fuses, comprising the following steps: S1. During normal operation, the edge computing module acquires the micro-disturbance tilt sequence of the fuse body's fuse element through a three-axis tilt sensor, and collects the transient voltage sequence of the energy storage capacitor on the secondary side of the CT power supply unit through a sampling circuit. S2. Extract phase space features based on the micro-perturbation tilt angle sequence, generate a mechanical stiffness index, and output fatigue warning information based on the mechanical stiffness index; S3. Perform differential operation based on the transient voltage sequence to obtain the voltage mutation factor, and continuously compare the voltage mutation factor with the mutation threshold; S4. When the voltage mutation factor reaches the mutation threshold, stop the listening task, perform boundary optimization based on the transient voltage sequence and thermal breakdown critical value, and obtain the limit flash time window. S5. Based on the closing of the control switch under the limit flash time window, the surge energy of the energy storage capacitor is injected into the LED light source, driving the LED light source to irradiate the hot colored plastic layer laid on the contact surface, and the detection module receives the red and blue reflected voltages reflected by the hot colored plastic layer. S6. Introduce the voltage mutation factor into the compensation model to correct the red and blue reflected voltages, calculate the color transfer index, write the color transfer index into the non-volatile memory for latching, and output a conduction signal to the bypass switch to short-circuit the secondary side of the CT power supply unit.

[0005] By adopting the above technical solution, transient energy hijacking drives the LED light source and introduces factor-corrected spectrum, thereby locking the true extreme thermal data before physical short circuit. This improves the problem that most traditional fuse monitoring devices use direct bypass discharge, which causes permanent loss of extreme thermal evidence when a fault occurs due to instantaneous power loss of the system.

[0006] Secondly, the present invention provides the following technical solution: an intelligent monitoring system for fuses, comprising the following modules: A fuse body, the fuse body comprising contacts and a fusible element, wherein the surface of the contacts is coated with a heat-colored plastic layer; A three-axis tilt sensor is fixed on the molten metal carrier to acquire the micro-disturbance tilt angle sequence of the molten metal carrier. A sampling circuit, which is connected to the energy storage capacitor, is used to acquire the transient voltage sequence of the energy storage capacitor; The CT power supply unit has the energy storage capacitor connected in parallel on its secondary side. The positive terminal of the energy storage capacitor is connected to the positive terminal of the LED light source through a first branch series control switch and to the drain of the bypass switch through a second branch. The source of the bypass switch is grounded. A detection module is positioned against the thermochromic plastic layer to receive the red and blue reflected voltages from the thermochromic plastic layer. Non-volatile memory, connected to the edge computing module, is used to latch color transfer indices; The edge computing module is electrically connected to the triaxial tilt sensor, the sampling loop, the control switch, the detection module, the non-volatile memory, and the bypass switch. The edge computing module internally includes: The background listening unit is used to schedule the micro-disturbance tilt sequence through the triaxial tilt sensor during normal operation, and synchronously transmit the transient voltage sequence through the sampling loop. The mechanical early warning unit is used to extract phase space features based on the micro-perturbation tilt angle sequence, generate a mechanical stiffness index, and output fatigue early warning information based on the mechanical stiffness index. A transient differential unit is used to perform differential operations based on the transient voltage sequence to obtain a voltage mutation factor, and continuously compare the voltage mutation factor with a mutation threshold. A boundary optimization unit is used to stop the listening task when the voltage mutation factor reaches the mutation threshold, and to perform optimization calculations based on the transient voltage sequence and the thermal breakdown critical value to obtain the limit flash time window. An energy hijacking drive unit is used to control the closing of the control switch based on the extreme flash time window, inject the surge energy of the energy storage capacitor into the LED light source to emit penetrating flashes to irradiate the hot-colored plastic layer, and simultaneously read the red and blue reflected voltages collected by the detection module; The self-healing latch unit is used to introduce the voltage mutation factor into the compensation model to correct the red and blue reflected voltages in order to calculate the color transfer index, write the color transfer index into the non-volatile memory for latching, and output a conduction signal to the bypass switch to physically short-circuit the secondary side of the CT power supply unit.

[0007] The present invention has the following beneficial effects: 1. In this invention, the LED light source is driven by transient energy hijacking and a factor-corrected spectrum is introduced. Then, the true extreme thermal data is locked before the physical short circuit. This improves the problem that most traditional fuse monitoring devices use direct bypass discharge, which causes permanent loss of extreme thermal evidence when the fault occurs due to the instantaneous power loss of the system.

[0008] 2. In this invention, mechanical stiffness index is generated by extracting phase space features based on micro-perturbation tilt angle sequence, thereby eliminating common mode displacement interference caused by the swing of the rod with the wind. This improves the problem that traditional fuses mostly use the absolute swing angle discrimination method, which cannot distinguish between increased wind force and loosening of the shaft, resulting in frequent false alarms in windy weather.

[0009] 3. In this invention, the abrupt change factor is obtained by performing differential operation based on transient voltage sequence and triggering boundary optimization. Then, the steep slope of energy change is intercepted before the capacitor breaks through the breakdown red line. This improves the problem that most traditional monitoring boards use static overvoltage thresholds, and the core circuit is instantly burned out in the initial stage of the surge due to the time delay of hardware sampling.

[0010] 4. In this invention, by exciting the LED light source to irradiate the hot-colored plastic layer and introducing a mutation factor to correct the reflected voltage, the spectral drift caused by the strong electric field is eliminated, thereby improving the problem that most traditional temperature measuring devices use low light intensity detection, and the weak light beam is easily blocked by the dirt layer, resulting in the distortion of temperature rise data in a highly polluted environment. Attached Figure Description

[0011] Figure 1 This is a flowchart of a fuse intelligent monitoring method proposed in this invention; Figure 2This is a flowchart of the mechanical stiffness index generation and fatigue early warning of a fuse intelligent monitoring method proposed in this invention. Figure 3 The flowchart below shows the voltage mutation factor detection and limit flash time window optimization of the intelligent monitoring method for fuses proposed in this invention. Figure 4 This is a flowchart of the energy hijacking drive and color transfer index latching process of the intelligent monitoring method for fuses proposed in this invention. Figure 5 This is an architecture diagram of an intelligent fuse monitoring system proposed in this invention. Detailed Implementation

[0012] The technical solutions in 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.

[0013] In the first embodiment of the present invention, a method and system for intelligent monitoring of fuses are provided, such as... Figures 1-2 As shown. To ensure the full disclosure and repeatability of the monitoring method and system technical features in this embodiment, the specifications and parameters of the underlying hardware carrier and core physical medium of this system are fixed as follows: The thermochromic plastic layer applied to the contact surface has a base resin made of a blend of high-pressure resistant polyvinylidene fluoride and polytetrafluoroethylene, with microcapsule thermochromic pigments doped with vanadium dioxide uniformly dispersed inside. This thermochromic plastic layer has reversible color-changing properties, with a color phase transition temperature range specified as 75 degrees Celsius to 85 degrees Celsius. It reversibly transitions from a pure blue state at a low temperature to a dark red state at a high temperature, with a medium thermal response time of less than 10 milliseconds.

[0014] The CT power-taking unit contains a miniature closed-type instrument transformer with a transformation ratio of 200 amperes to 1 ampere and a rated output secondary current of 1 ampere. The energy storage capacitor is a high-frequency, low-impedance electrolytic capacitor with an inherent capacitance of 470 microfarads, a withstand voltage of 450 volts, and an equivalent series internal resistance of less than 15 milliohms. The extreme constant of the safe voltage charging slope is specifically assigned a value of 35 volts per second, based on the physical limit slope of charging the 470 microfarad capacitor with a full-load current of 1 ampere on the secondary side of the instrument transformer.

[0015] The thermal breakdown threshold is set at 4.5 joules. This value is strictly anchored to the junction temperature thermal breakdown red line of the metal-oxide-semiconductor field-effect transistor inside the main control buck regulator chip on the local monitoring board. That is, absorbing 4.5 joules of heat instantly will cause the junction temperature to exceed 150 degrees Celsius and cause irreversible vaporization damage.

[0016] The excitation charge diffusion constant is explicitly assigned a value of 0.042 to characterize the inherent quantum confinement drift slope of the internal carrier recombination center of the inorganic indium gallium nitride light-emitting diode to broaden the high-frequency bandgap under transient overload current driving.

[0017] The contamination layer covering the contact surface is defined as a mixed contamination layer of heavy salt spray and carbon black from coastal industrial environments. Its bidirectional transmittance and reflectance factors for 650 nm red light and 450 nm blue light fall within the dimensionless scalar ranges of 0.15 to 0.35 and 0.05 to 0.12, respectively. The specific monitoring steps in this embodiment include: S1. During normal operation, the edge computing module acquires the micro-disturbance tilt sequence of the fuse body's fuse element through a three-axis tilt sensor, and collects the transient voltage sequence of the energy storage capacitor on the secondary side of the CT power supply unit through a sampling circuit. Furthermore, in S1, during normal operation, the edge computing module acquires the micro-disturbance tilt sequence of the fuse body's fusible element through a triaxial tilt sensor, and collects the transient voltage sequence of the secondary side energy storage capacitor of the CT power extraction unit through a sampling circuit, including: The hardware communication interface of the triaxial tilt sensor is activated, and the triaxial acceleration components are continuously read according to the set sampling clock cycle. The read triaxial acceleration components are then transformed by the coordinate rotation matrix to extract the micro-perturbation tilt sequence that characterizes the deviation of the vertical axis of the molten workpiece. The high-impedance voltage divider branch in the sampling circuit is turned on to synchronously capture the analog levels across the energy storage capacitor. The captured analog levels are then discretized and calculated by the true RMS conversion module to generate a transient voltage sequence.

[0018] Specifically, the input terminal intercepts the simulated original gravitational acceleration signal of the fuse carrier element, excited by the alternating natural wind field and the 50 Hz power frequency electromagnetic Lorentz force. The hardware communication interface sequentially reads the discrete digital components according to a fixed sampling clock cycle of 0.01 seconds. The edge computing module calls the built-in gravity vector projection formula to perform arctangent rotation matrix transformation, and the output terminal continuously pushes a sequence of micro-perturbation tilt angles characterizing the relative deviation of the fuse carrier element's vertical axis.

[0019] The mathematical model and parameter definitions are as follows: During the initial static initialization phase when the monitoring device is first powered on, the initial gravitational acceleration vector output by the triaxial attitude tilt sensor is extracted. The static pitch angle of the fuse mounting base caused by the tilt of the utility pole was calculated. With static roll angle : ; ; During normal operation, the first Real-time triaxial acceleration vector read per sampling clock cycle Left multiplied by the static pitch angle With static roll angle Constructed coordinate inverse rotation matrix By stripping the static installation tilt angle of the utility pole, the pure dynamic acceleration components in the coordinate system of the load-bearing component are obtained. : ; Based on the pure dynamic acceleration components, an arctangent projection solution is performed to generate a sequence of micro-perturbation tilt angles: ; in These represent the initial gravitational components measured along the three axes of the sensor under static conditions, in meters per square second. The static spatial deflection angle of the base, expressed in radians; Characterizes the real-time dynamic acceleration components after coordinate system leveling and restoration, in meters per square second; Characterizing the first The micro-disturbance tilt angle value of the oscillation of the carrier around the base pin is calculated and output every sampling clock cycle, in degrees.

[0020] When drop-out fuses are subjected to long-term wind loads outdoors, the fixed base and the utility pole body will experience overall low-frequency oscillation. If the original acceleration values ​​are directly used as the basis for judging mechanical jamming, the background displacement of the base's swaying will be misjudged as loose contacts. This step uses a triaxial gravity tangent rotation matrix to completely isolate the common-mode displacement vector of the base's overall swaying with the environment at the algebraic level, accurately extracting the pure attitude variable of the fuse-carrying component's slight relative dry friction oscillation around the base pin axis. This variable establishes a clean zero-point data benchmark for subsequent determination of whether the hinge's mechanical stiffness has entered the nonlinear region of rusting.

[0021] The input terminal receives the original power frequency AC voltage, which is filled with high-frequency spikes and glitches, across the energy storage capacitor on the secondary side of the CT power supply unit. A high-impedance voltage divider branch feeds a low-voltage analog signal stream into the analog-to-digital conversion channel at a constant attenuation ratio. The edge computing module calls the discretized true RMS equation to perform arithmetic reconstruction on the analog signal stream, and the output terminal periodically generates a smooth transient voltage sequence.

[0022] The mathematical model and parameter definitions are as follows: ; in Characterizing the first The transient voltage value is calculated and output by a true RMS calculation window, with the physical dimension being volts. The cumulative total of discrete voltage sampling points within a single calculation window is represented by a fixed value of 128, based on the 50 Hz distribution network benchmark in my country. Characterizing the first [value] intercepted in microsecond steps via the high-impedance voltage divider branch. A number of instantaneous simulated voltage divider samples, with the physical dimension in volts.

[0023] When the CT power supply circuit encounters a sudden change in distribution network load or a minor disturbance in the line, the instantaneous charging level of the energy storage capacitor will generate extremely high non-sinusoidal spike pulses. If the edge computing module directly collects the peak voltage for overvoltage limit judgment, the system will experience false triggering of high-frequency protection. This step prevents the bypass discharge of the energy stored in the energy storage capacitor by introducing a megaohm-level high-impedance branch, and uses true RMS root-based calculation to smooth the integral of the charging and discharging spikes into a voltage reference that is strictly equivalent to the actual accumulated Joule heat of the capacitor, ensuring that the voltage mutation factor calculated by the subsequent time-domain differential is absolutely immune to algebraic interference from high-frequency harmonic spikes.

[0024] By configuring the aforementioned attitude calculation and true RMS value calculation in parallel within the normal operation task stack, independent sampling channels for microscopic mechanical parameters and macroscopic energy parameters are constructed at the underlying logic level. The edge computing module only needs to maintain extremely low-power polling of the hardware interface to provide deterministic linear data support for subsequent identification of mechanical corrosion dead zones of the shaft without increasing the material overhead of additional detection devices, and to ensure the absolute purity of the energy function determination boundary when encountering severe short-circuit surges.

[0025] S2. Extract phase space features based on micro-perturbation tilt angle sequence, generate mechanical stiffness index, and output fatigue early warning information based on mechanical stiffness index; Furthermore, in S2, phase space features are extracted based on the micro-perturbation tilt angle sequence to generate mechanical stiffness indices, including: Extract the absolute value of the tilt angle difference between adjacent sampling points in the micro-perturbation tilt angle sequence, and calculate the arithmetic mean to obtain the absolute motion trajectory evolution rate; Calculate the root mean square amplitude of the in-time domain of the micro-perturbation tilt angle sequence to obtain the potential energy fluctuation of the system; The quotient is obtained by dividing the absolute motion trajectory evolution rate by the system potential energy fluctuation, and then multiplying the quotient by the time constant correction benchmark to generate the dimensionless mechanical stiffness index.

[0026] In S2, fatigue warning information is output based on the mechanical stiffness index, including: Extract the factory mechanical stiffness reference value pre-stored in the storage unit; Compare the mechanical stiffness index with the factory mechanical stiffness benchmark value; If the comparison result indicates that the mechanical stiffness index is greater than the factory mechanical stiffness benchmark value of the first set ratio, fatigue warning information of dry friction jamming of the shaft will be output. If the comparison result indicates that the mechanical stiffness index is less than the factory mechanical stiffness benchmark value of the second set ratio, fatigue warning information of softening of the return spring stiffness will be output.

[0027] Specifically, the input receives a discrete sequence of clean, micro-perturbation tilt angles of the molten metal carrier pushed by the pre-filtering channel, and simultaneously reads the fixed factory mechanical stiffness reference value from the physical address block specified in the non-volatile memory. The arithmetic logic unit inside the edge computing module executes first-order difference, absolute value summation, root mean square root extraction, and algebraic division instructions sequentially. After the calculation is complete, the output directly pushes a dimensionless real-time mechanical stiffness index value to the warning register; when this value is determined to have exceeded the set proportional threshold boundary, the output pushes the corresponding mechanical jamming or spring fatigue code to the external status indication bus via a hardware interrupt.

[0028] The underlying computational model for generating the mechanical stiffness index is configured as follows: ; in The dimensionless mechanical stiffness index represents the current solution cycle. The cumulative total of discrete tilt angle sampling points within a single calculation window is represented by a fixed value of 512, based on the cache depth of the edge computing module. Characterizing the first The tilt angle values ​​of micro-perturbations captured at discrete sampling times, with the physical dimension being degrees. It represents the tilt angle value of the micro-perturbation intercepted at the immediately preceding discrete sampling time, with the physical dimension being degrees. It represents the arithmetic mean of the tilt angles of all sampled points within the calculation window, with the physical dimension being degrees. The constant hardware sampling clock step size between adjacent sampling points is represented by a physical unit of seconds. The time constant correction benchmark, which is embedded within the edge computing module, is in seconds and is used to cancel out the reciprocal second dimension contained in the quotient into a pure algebraic scalar. Here, it is explicitly set to a constant of 0.25 seconds.

[0029] The fatigue early warning unit executes a two-way threshold criterion formula group configuration for numerical comparison. The judgment condition for dry friction jamming of the rotating shaft is: ; The criterion for determining whether the reset spring has softened is: ; in The initial mechanical stiffness reference value entered into the non-volatile memory during the factory inspection stage is dimensionless. The first set proportionality constant is dimensionless, and its value is explicitly fixed at 1.5. The second set proportionality constant is dimensionless, and here it is explicitly set to a constant of 0.65. Based on the principle of mechanical equivalence, the dimensionless mechanical stiffness exponent is calculated. The actual physical torsional stiffness of the fuse pin follows a linear mapping equation, with the factory stiffness calibration coefficient explicitly assigned a value of 18.5 Nm / radian. The number of sampling points in the formula is fixed at 512, representing the standard integer depth of the direct memory access buffer built into the edge computing module; the time constant is assigned a value of 0.25 seconds, representing one-quarter of the empirical benchmark for capturing a single complete viscous slip oscillation cycle of the fuse element under natural wind conditions. In the bidirectional threshold for fatigue warning, the first setting ratio of 1.5 times and the second setting ratio of 0.65 times the factory benchmark are statistically critical inflection points obtained from performing 5000 full-life-cycle mechanical drop-overturn fatigue life tests on the standard fuse body. For fuses of different pitch models, the benchmark value is proportional to the main shaft radius. Linear scaling with the same proportion is sufficient.

[0030] The closing holding force and drop-off trajectory of a drop-out fuse are governed by the underlying nonlinear Da Vinci dynamics. Under the influence of random strong winds and alternating excitation of power frequency electromagnetic force outdoors, the fuse carrier in the closed state will experience extremely small random mechanical swaying around the base pin.

[0031] If only the apparent absolute amplitude of the micro-shaking of the fuse carrier is monitored, the monitoring system cannot determine whether the current severe shaking is due to increased external wind force or mechanical loosening of the hinge. In this refinement step, the first-order difference arithmetic mean of the numerator term characterizes the instantaneous discrete angular velocity manifold of the fuse carrier's oscillation, while the root-mean-square amplitude in the time domain of the denominator term characterizes the static position potential energy envelope of the oscillation. Dividing the two within the arithmetic logic unit is physically equivalent to extracting the average viscous slip rate of the fuse carrier's micro-motion trajectory in a two-dimensional pseudo-phase space plane with the oscillation angle-angular velocity as the orthogonal axis. When the fuse base pin hardware is subjected to long-term outdoor rainwater erosion and develops severe rust, the shaft deteriorates from viscous friction to dry friction. The viscous slip effect of dry friction forces the phase space trajectory to undergo high-frequency twitching distortion, thereby instantaneously increasing the value of the numerator term in the formula. The frictional properties of the shaft deteriorate from smooth viscous lubricating friction to dry friction. The unique viscous slip effect of dry friction forces the molten material to produce extremely small but extremely high-frequency reciprocating twitching oscillations under wind power. This high-frequency micro-oscillation instantly increases the sum of the absolute values ​​of the first-order difference of the tilt angle, causing a step-like increase in the value of the numerator term in the formula.

[0032] The root-mean-square amplitude introduced in the denominator directly characterizes the total static potential energy envelope of the molten component deviating from the vertical line of gravity. Dividing the numerator of the high-frequency angular velocity directly by the denominator of the total amplitude establishes a natural excitation adaptive normalization mechanism at the algebraic level. When the external natural wind force increases exponentially, the numerator angular velocity and the denominator amplitude will amplify synchronously in the same proportion; their division directly eliminates the random variable of the natural wind force. The calculated quotient is purified and is only controlled by the stiffness and damping degradation state of the mechanical hinge itself.

[0033] The bidirectional threshold comparison mechanism establishes a simplified classification criterion for mechanical nonlinear degradation. When the calculated stiffness index exceeds 1.5 times the factory standard, the system can identify the physical fact that the hinge has rusted and jammed before the fuse blows. When the calculated stiffness index falls below 0.65 times the factory standard, it accurately reflects the softening characteristics of the return spring due to metal fatigue caused by years of stress, resulting in tension attenuation and sluggish oscillation of the fuse-carrying component. This design eliminates the hardware computational burden of adding a high-frequency force gauge and a microprocessor to perform frequency domain fast Fourier transform. It establishes a closed-loop basis for early warning of hidden mechanical faults in fuses by relying solely on one-dimensional tilt angle time-domain algebraic dimensionality reduction.

[0034] like Figure 3 As shown, S3, perform differential operation based on transient voltage sequence to obtain voltage mutation factor, and continuously compare voltage mutation factor with mutation threshold; Furthermore, in S3, differential operations are performed based on the transient voltage sequence to obtain the voltage change factor, including: Extract the transient voltage sequences of two adjacent differential time steps, perform a subtraction operation and divide by the corresponding differential time step to obtain the actual voltage change rate; Extract the preset extreme constant of the charging slope at the safe voltage; Calculate the ratio of the actual voltage change rate to the extreme constant of the safe voltage charging slope, and multiply the ratio by the proportional linkage coefficient to generate a dimensionless voltage mutation factor.

[0035] Specifically, the input continuously receives the discrete sequence of transient effective voltage values ​​pushed by the preceding true effective value conversion channel according to the calculation cycle. The hardware multiply-accumulate unit inside the edge computing module sequentially calls the historical cache values ​​of the current clock window and the previous clock window, and executes the first-order backward differential instruction and fixed-point division instruction in sequence. The calculated real-time scalar is directly overwritten into the transient mutation factor working register. The hardware comparator continuously reads the value of this register and compares it with the mutation threshold red line in an infinite loop; once the limit is exceeded, the output immediately cuts off the current differential polling process and sends a high-level hardware interrupt pulse to the system kernel bus to start short-circuit boundary optimization.

[0036] The underlying algebraic model configuration for generating the voltage mutation factor is as follows: ; in The dimensionless transient voltage overload mutation rate factor characterizes the output of the current computation step size. The value represents the real-time voltage value captured by the current true RMS calculation window, with the physical dimension in volts. The value represents the historical voltage value captured by the immediately preceding true RMS calculation window, with the physical dimension in volts. The fixed differential time step is used to characterize the push points of two adjacent true effective value sequences. Based on the half-wave rectification period of the AC power in the 50 Hz distribution network, the fixed value here is a constant of 0.01 seconds. The extreme value constant of the safe voltage charging slope, pre-installed in the non-volatile memory, is expressed in volts per second. This constant maps the physical limit boundary of the natural voltage rise per second across the secondary energy storage capacitor of the CT self-driven power unit under the maximum permissible full load condition. The proportional linkage coefficient is dimensionless and is used to map the physical ratio calculated by division into an engineering integer range scalar that is easy for microprocessors to process. Here, it is explicitly set to a constant 1.

[0037] The trigger criterion formula for the continuous comparison logic is configured as follows: ; in The preset mutation threshold, dimensionless, is explicitly fixed at a constant of 10. This threshold originates from the boundary condition calculation of the transient simulation of the entire short-circuit process in a 36 kV distribution network: when a conventional large load is instantaneously connected or disconnected in the distribution network, the slope of the surge in induced energy on the secondary side cannot exceed 8 times the charging limit constant due to the clamping effect of the bus inductance; however, when a metallic phase-to-phase short circuit occurs, the derivative of the initial impulse of the short circuit instantly jumps to more than 12 times the design limit. The constant 10 is precisely located in the pure isolation zone between the upper limit of the conventional grid disturbance potential energy and the lower limit of the metallic short-circuit energy injection, objectively constructing an absolute threshold for fault identification in the entire network.

[0038] When a routine large load is connected or disconnected in a 36 kV medium-voltage distribution network where a drop-out fuse is located, the energy increment fed into the secondary side of the current transformer (CT) on the busbar exhibits a low-slope gradual change characteristic. When the distribution network encounters lightning overvoltage or a phase-to-phase severe metallic short circuit, the large incoming current will force the Joule heat injected across the energy storage capacitor to rise explosively and steeply within milliseconds.

[0039] Simply relying on setting an absolute voltage threshold to trigger bypass protection will lead to an engineering deadlock: if the threshold is set too low, normal load fluctuations in the power grid will frequently cause the monitoring system to malfunction and short-circuit power outages; if the threshold is set too high, by the time the microcontroller's ADC channel actually reads an excessively high voltage, the violent surge spikes will have already completely burned through the microcontroller pins and the back-end photoelectric detection chip.

[0040] This step extracts the first-order time derivative of the charging potential energy of the energy storage capacitor by subtracting adjacent effective values ​​and dividing by the time step, thus mathematically capturing this first-order time derivative. This actual charging derivative is then directly divided by the extreme value constant of the charging slope under normal operating conditions, forcibly normalizing the overvoltage value, which was originally highly dependent on the specific bus voltage reference, into a dimensionless abrupt change factor that purely characterizes the steepness of the surge rise. This reduction operation completely eliminates common-mode background interference caused by static deviations in the absolute value of the secondary-side induced voltage due to inconsistent transformer capacities in different distribution network areas.

[0041] The continuous comparison mechanism establishes an extremely sensitive event-driven threshold. As long as the mutation factor is below a constant of 10, the microprocessor core determines that the power distribution network is in a normal energy throughput oscillation range and maintains a differential polling state with extremely low power consumption. Once the mutation factor exceeds a constant of 10, it provides solid evidence at the algebraic level that the secondary side is suffering from a malignant high-frequency energy injection ten times the design limit. This differential mechanism captures the deterministic physical slope precursor of a short-circuit fault in the microsecond-level initial stage before the absolute voltage of the energy storage capacitor crosses the hardware breakdown threshold, thus securing an important nanosecond-level time window for the subsequent S4 preemptive execution of LED high-intensity light optimization.

[0042] like Figure 3 As shown, S4, when the voltage mutation factor reaches the mutation threshold, the listening task is stopped, and boundary optimization is performed based on the transient voltage sequence and the thermal breakdown critical value to obtain the limit flash time window; Furthermore, in S4, boundary optimization is performed based on the transient voltage sequence and the thermal breakdown critical value to obtain the limiting flash time window, including: Based on the transient voltage sequence at the current sampling time and the inherent capacitance value of the energy storage capacitor, the dynamic energy function characterizing the accumulated energy inside the energy storage capacitor is reconstructed. Calculate the energy difference between the thermal breakdown critical value and the dynamic energy function to obtain the remaining safe injection energy; Extract the rated power of the LED light source, calculate the ratio of the remaining safe injected energy to the product of the voltage mutation factor and the rated power, and use the calculated ratio as the limiting flash time window when the energy-time derivative is in a negative constant state.

[0043] Specifically, the input intercepts the over-limit trigger pulse output from the preceding differential logic stage, and simultaneously retrieves the transient effective value voltage buffer sequence, the preset thermal breakdown threshold value in the non-volatile memory, and the factory rated operating power of the LED light source. The multiply-accumulate unit inside the edge computing module sequentially executes squaring, scalar multiplication, algebraic subtraction, and fixed-point division instructions. After the calculation is completed, the output sends the obtained limit flash time window scalar value to the pulse width modulation generator, which directly locks the conduction maintenance cycle of the high-frequency electronic switch.

[0044] The underlying formula configuration for the reconstructed dynamic energy function of the energy storage capacitor is as follows: ; in This represents the actual accumulated joules of thermophysical energy inside the energy storage capacitor at the current sampling moment, measured in joules. The inherent capacitance value of an energy storage capacitor, expressed in farads. It represents the effective value of the transient AC voltage intercepted by the sampling circuit at the current moment, and the unit is volts.

[0045] The algebraic subtraction formula for calculating the remaining safe injected energy is configured as follows: ; in Characterizes the remaining safe thermal tolerance of the system at the current distance from irreversible thermal collapse, expressed in joules. The thermal breakdown critical value, in joules, characterizes the overall thermal collapse damage of the low-voltage components of the system. According to the 36 kV medium-voltage fitting inspection standard, the value here is explicitly set to a constant of 4.5 joules.

[0046] The optimization model configuration for solving the limiting flash time window is as follows: ; in The limit flash time window, in seconds, represents the output of the edge computing module's optimization calculation. The remaining safe thermal capacity of a capacitor from its current thermal breakdown threshold, expressed in joules. Characterizes the voltage mutation factor pushed in real time; The rated operating power of a light-emitting diode (LED) light source, expressed in watts. The transient overload power amplification factor is dimensionless and is used to characterize the peak power of an LED light source that is forced to exceed its rated power by the short-circuit surge current under the physical direct discharge of the control switch. Based on the 36 kV short-circuit impact test benchmark data, the fixed value is defined as a constant of 12.

[0047] When a distribution network experiences a severe phase-to-phase short circuit, the surge in short-circuit current on the incoming bus can inject tens of times more energy than the design limit into the current transformer (CT) power supply circuit within milliseconds. Conventional intelligent monitoring devices face an engineering paradox under this condition: bypassing the short circuit for self-protection results in the complete loss of extreme contact temperature rise data at the initial stage of the fault due to instantaneous power loss; conversely, forcibly maintaining power supply sampling would cause the low-voltage core board to be instantly punctured and burned out by the high voltage.

[0048] This refinement step constructs a Lyapunov boundary optimization model based on real-time Joule heat accumulation of capacitors, reconstructing destructive surge overvoltages at the physical level as transient high-energy excitation sources that pierce through the surface contamination of equipment. The edge computing module no longer executes rigid delayed bypass cycles, but instead directly participates in the optimization process using the intensity of the grid surge as a dynamic variable in the denominator. When the inrush short-circuit current slope is extremely steep, the voltage mutation factor increases dramatically, forcing the calculated flash time window to shrink significantly to the microsecond level; when the fault current slope is relatively gentle, this time window naturally extends backward.

[0049] This algebraic solution mechanism establishes the absolute critical point at which the system's energy time derivative transitions from positive definite to negative definite. Within the extremely short-second interval before the capacitor's accumulated heat reaches the thermal breakdown threshold, the edge computing module precisely extracts surge potential energy to drive the light source to unleash an overload of intense light. This action not only utilizes the last available electrical energy before the short circuit to pierce the heavily contaminated layer covering the contact surface, but also ensures, through a perfectly precise pulse width cutoff timing, that the semiconductor junction temperature is forcibly reduced to zero before reaching the physical damage boundary. This establishes a mechanism for the coexistence of fixed extreme values ​​of temperature rise data and overall hardware stability under harsh operating conditions.

[0050] like Figure 4 As shown, S5 controls the switch to close based on the limit flash time window, injects the surge energy of the energy storage capacitor into the LED light source, drives the LED light source to irradiate the hot colored plastic layer laid on the contact surface, and obtains the red and blue reflected voltages reflected by the hot colored plastic layer received by the detection module. Furthermore, in S5, based on the limit flash time window control switch closure, the surge energy of the energy storage capacitor is injected into the LED light source, driving the LED light source to irradiate the thermoplastic layer applied to the contact surface, including... Under the condition that the starting moment of the limit flash time window is reached, the trigger control switch switches from the open state to the closed state; The surge energy in the energy storage capacitor, which increases logarithmically, bypasses the voltage regulation and buck circuit and is directly input to the positive terminal of the LED light source; The transient overload current exceeding the rated operating power is used to excite the LED light source to emit a transient penetrating flash, which illuminates the hot-colored plastic layer covered by the target dirt.

[0051] Specifically, this step introduces the timing pulse signal of the limit flash time window, determined by optimization in the previous step S4, as the hard trigger condition for the control switch state toggling. The hardware pin of the edge computing module outputs a saturated drive level, and the control switch response level switches from the open state to the closed state, instantly opening a non-inductive, low-impedance discharge path on the electrical road that avoids the weak current voltage regulation and step-down circuit. The transient large current in the energy storage capacitor is directly injected into the positive terminal of the LED light source along the discharge path. At the output end, the first narrowband red light receiving diode and the second narrowband blue light receiving diode inside the detection module capture the original red reflected voltage and the original blue reflected voltage in the form of nanosecond-level spike pulses in real time.

[0052] At the instant of direct overload injection of transient surge energy, the original red reflected voltage and the original blue reflected voltage, which are synchronously received and output by the first narrowband red light receiving diode and the second narrowband blue light receiving diode inside the detection module, respectively follow the following optical physical mathematical models: ; ; in and The original red reflected voltage and the original blue reflected voltage are represented respectively, in volts; The overall gain coefficient characterizing the photoelectric receiving circuit, expressed in volts per ampere; Characterizes the transient level of the energy storage capacitor at the moment of flash initiation, measured in volts; Characterizes the total internal resistance of the direct-connected discharge circuit, in ohms; and The dimensionless bidirectional penetration attenuation factor of a heavily contaminated layer on the contact surface for 650 nm red light and 450 nm blue light was characterized.

[0053] From a physical perspective, since the geometric particle size of the silt and oxide particles within the target contamination layer is on the same order of magnitude as the wavelength of visible light, photons are dominated by Mie scattering when penetrating the contaminated medium. Because the Mie scattering cross-section increases dramatically with decreasing wavelength, the scattering loss encountered by short-wavelength blue light when penetrating the contamination layer is far greater than that of long-wavelength red light, objectively and strictly satisfying the following identity inequality: ; Conventional drop-out fuse intelligent monitoring devices are typically equipped with DC voltage regulator chips to supply a constant current at the microwatt level to maintain the linear emission state of the light-emitting diodes. This steady-state low-emissivity light intensity is easily blocked by the industrial dust, silt, or oxide contaminants that accumulate on the surface of outdoor fittings in high-voltage power distribution networks over the years, causing the reflected light signal to be completely submerged in the environmental background noise.

[0054] This step breaks with the conventional design bias that power distribution network monitoring devices must maintain voltage stability. It uses a control switch to forcefully guide surge energy, which increases logarithmically, to bypass all weak voltage regulation and step-down circuits. This open-circuit hard connection directly injects several joules of residual short-circuit surge potential energy into the positive electrode of the LED light source without reservation. By constructing this microsecond-level extreme overload current, it forces instantaneous, extremely high-density carrier recombination within the energy bands of the LED semiconductor.

[0055] Within a microsecond window before vaporization and destruction, the overload-excited LED light source emits a transient high-energy penetrating flash, exceeding its conventional rated operating power by tens of times. This high-photon-flux flash forcefully pierces the heavily contaminated surface layer, reaching the underlying hot-colored plastic layer. The reflected beam, carrying the color characteristics of the contact's real-time temperature rise, then penetrates the contamination layer again and is ultimately captured by a narrow-band receiving diode. This electrical topology avoids the conventional approach of increasing the physical size of the light-emitting device, utilizing short-circuit power compression within an extremely short time window to achieve non-destructive extraction of weak optical parameters under extremely harsh contamination conditions.

[0056] like Figure 4 As shown, S6, the voltage mutation factor is introduced into the compensation model to correct the red and blue reflected voltages, the color transfer index is calculated, the color transfer index is written into the non-volatile memory for latching, and a conduction signal is output to the bypass switch to short-circuit the secondary side of the CT power supply unit. Furthermore, in S6, a voltage mutation factor is introduced into the compensation model to correct the red and blue reflection voltages, and the calculated color transfer index includes: Extract the excitation charge diffusion constant, which characterizes the band drift under overload injection, and calculate the product of the natural logarithm of the voltage mutation factor and the excitation charge diffusion constant to generate the dynamic compensation base. The red reflected voltage is multiplied by the correction coefficient after the dynamic compensation base is superimposed with a constant to obtain the corrected red reflected component, while the blue reflected voltage is kept as the original component. The dimensionless color transfer index is generated by performing a ratio division operation by subtracting the corrected red reflection component from the blue reflection voltage as the numerator and adding the corrected red reflection component to the blue reflection voltage as the denominator.

[0057] In S6, writing the color transfer index into the non-volatile memory latch and outputting an on signal to the bypass switch to short-circuit the secondary side of the CT power supply unit includes: Enable the direct memory access channel inside the edge computing module and write the color transfer index into the physical address block specified by the non-volatile memory; In the machine cycle that detects that the write completion flag returned by the direct memory access channel is set, a low cutoff level is output to the control switch to cut off the power supply circuit of the LED light source. The saturated conduction level signal is output synchronously to the control electrode of the bypass switch, and the subsequent surge energy generated by the secondary side of the CT power supply unit is discharged to the ground terminal through the drain-source channel.

[0058] Specifically, the input terminal synchronously introduces the original red and blue reflection voltages intercepted by the detection module, and retrieves the voltage mutation factor calculated in real time in step S3. The arithmetic logic unit inside the edge computing module extracts the preset excitation charge diffusion constant from the register, and sequentially executes natural logarithm calculation, multiplication scaling, and algebraic summation instructions to reconstruct the corrected red reflection component. Subsequently, the arithmetic logic unit performs standard ratio division, and the output terminal directly pushes the calculated dimensionless color transfer index to the direct memory access controller.

[0059] The algebraic compensation model for the red reflection component is configured as follows: ; The final solution model for the color transfer index is configured as follows: ; in The true limiting color transfer index, which is dimensionless, is represented by the output after inverse frequency shift correction. Characterized by the original blue reflected voltage captured and output by a narrow-band blue light receiving diode within the limiting flash time window, measured in volts. Characterized by the original red reflected voltage captured and output by a narrowband red light receiving diode within the limiting flash time window, measured in volts. Characterized by the red reflection component restored after joint correction of the excitation charge diffusion constant and the abrupt slope, in volts. Characterizes the dimensionless voltage mutation factor pushed in real time by the differential calculation branch. The dimensionless excitation charge diffusion constant characterizes the excitation charge diffusion rate pre-placed within the storage cell. This constant maps the inherent physical drift slope of the exciton recombination center towards the short-wavelength direction when the semiconductor band inside the microfluidic LED light source is subjected to a strong electric field injection far exceeding the rated power.

[0060] When a light-emitting diode (LED) encounters a short-circuit transient overload current, the carrier injection density within the PN junction exhibits an explosive increase, and the band structure, under the pressure of a strong electric field, produces a significant quantum confinement Stark effect. At the semiconductor quantum physics level, this effect causes the frequency shift of the emission peak wavelength to be strictly proportional to the natural logarithm of the injected forward current density, following the empirical law of bandgap deformation. Since the voltage jump factor calculated in this invention strictly linearly maps the short-circuit current derivative of the injected energy storage capacitor, the natural logarithm of the jump factor is introduced as an algebraic multiplication factor, which is the only isomorphic function for reconstructing the bandgap deformation topology within the PN junction. In the calculation architecture, only the red channel is compensated while the blue channel remains constant, based on a static reference design using a dual-color medium: the background reflected light of the thermochromic plastic layer at normal low temperatures is 450 nm pure blue light, and the absorption rate of this band remains constant throughout the material phase transition, serving as the static denominator anchor point for the system calculation; while the 650 nm red light reflectivity directly maps the dynamic bandgap width generated by the vanadium dioxide phase transition. Therefore, by locking the blue light as a static reference and performing frequency shift inverse logarithmic compensation only on the red light component, the electric field distortion error can be completely removed at the algebraic level, thus purifying the true limit color transfer index.

[0061] At the system hardware communication layer, the edge computing module establishes an electrical connection with the triaxial tilt sensor via a 400 kHz high-speed internal integrated circuit bus; it connects to the non-volatile memory via a 20 MHz serial peripheral interface bus; and it uses a parallel hardware first-in-first-out queue interface for data scheduling with the voltage sampling circuit. The input receives the true limit color transfer index output by the arithmetic logic unit. The direct memory access channel responds to the write enable signal, forcibly bypassing the CPU's internal register stack and the regular program scheduling queue, and directly hard-writes the index to the specified physical address block within four machine clock cycles. This invention utilizes a transient overload current exceeding 12 times the rated operating power to drive a light-emitting diode to emit a transient penetrating flash. The physical reliability of the device, which prevents it from crashing, is controlled by a thermal capacity time constant interlocking mechanism: the physical thermal conduction expansion time constant of the bare die of the light-emitting diode is about 2 milliseconds, while the limit flash time window calculated by this invention is strictly clamped within 50 microseconds. Before the high heat generated by the overload current is conducted from the PN junction to the gold wire pads, the control switch is forcibly turned off, realizing the preemptive completion of the photon emission action before the adiabatic temperature rise critical point. During power-off self-protection, to avoid the risk of burn-through caused by the main control board losing power before the light source power supply is cut off due to the bypass switch being turned on first, the system introduces nanosecond-level dead-zone interlocking timing: After detecting the write completion hardware flag, the core sequencer of the edge computing module forces the output of a falling edge pulse with a slope of less than 50 nanoseconds to completely turn off the control switch; after a 2.5 microsecond absolute dead-zone isolation time, a saturation conduction high level is simultaneously poured into the gate drive channel of the high-power solid-state electronic bypass switch, completely physically short-circuiting the secondary side of the CT power supply unit. The subsequent thousands of amperes of destructive electrical energy generated by the bus short circuit are directly discharged back to the distribution network grounding terminal through the extremely low internal resistance drain-source channel. The power supply port of the local monitoring board is forced to be clamped to zero potential, and the core components as a whole enter a power-off physical dormant self-protection state. Afterwards, maintenance personnel can completely trace the disaster curve by reading the non-volatile memory.

[0062] Example 2: The HRW12-36 / 200A type drop-out fuse body is installed on the outdoor overhead lines of the 36 kV power distribution network located in a heavily polluted coastal industrial park. Under the long-term alternating excitation of strong winds and high salt spray corrosion, the flip pin of the fuse carrier is hidden and rusted, and the contacts experience abnormal temperature rise over a long period of time. When a sudden phase-to-phase metallic malignant short circuit occurs in a branch line of the distribution network, the large short-circuit current of the incoming line injects a highly destructive transient high-voltage surge into the secondary side of the CT self-driven power supply unit within milliseconds. At the moment of the short-circuit impact, the local monitoring system faces serious hardware and software coordination problems: if the edge computing module immediately conducts the bypass switch to protect itself from power loss, the extreme thermal degradation evidence on the contact surface at the moment of the short circuit will be permanently lost; if the power supply is forcibly maintained to drive the light source to penetrate the heavily contaminated layer on the contact surface for evidence collection, the violent surge will not only instantly physically vaporize the photoelectric semiconductor, but the extreme strong electric field generated by the overload transient current will also cause the light source energy band to undergo a Stark effect spectral blue shift, resulting in a severely distorted color shift index output by the dual-frequency detection circuit, causing the on-site investigation of power grid faults to fall into a double blind zone of hardware damage and data distortion. To solve the above problems, this invention provides an intelligent fuse monitoring system, the structure of which is as follows: Figure 5 As shown. The specific implementation process of this system is as follows: The edge computing module, relying on a triaxial tilt sensor and a background monitoring unit, directly converts the micro-disturbances of the load-bearing component under wind load into an evaluation of the shaft stiffness, eliminating the material overhead of external force measuring devices. When encountering a short circuit impact from the power distribution network, the transient differential and boundary optimization unit takes over control, forcibly cuts off polling and defines a safe time window. The energy hijacking unit drives a switch to directly inject the destructive surge potential energy in the capacitor through the voltage regulator circuit into the LED light source, exciting a transient high-penetration flash that pierces the contaminant layer on the contact surface for optical evidence collection. Subsequently, the self-healing latch unit introduces a voltage mutation factor to inversely correct the frequency shift error of the semiconductor's strong electric field, hard writes the calculated pure hot-state color index into the non-volatile memory, and drives the bypass switch to completely short-circuit the CT secondary side. The entire system completes a closed loop within microseconds, from pre-disaster mechanical early warning, to strong light penetration evidence collection during disasters, to post-disaster hardware power-off self-protection.

[0063] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of fuses, characterized in that, Includes the following steps: S1. During normal operation, the edge computing module acquires the micro-disturbance tilt sequence of the fuse body's fuse element through a three-axis tilt sensor, and collects the transient voltage sequence of the energy storage capacitor on the secondary side of the CT power supply unit through a sampling circuit. S2. Extract phase space features based on the micro-perturbation tilt angle sequence, generate a mechanical stiffness index, and output fatigue warning information based on the mechanical stiffness index; S3. Perform differential operation based on the transient voltage sequence to obtain the voltage mutation factor, and continuously compare the voltage mutation factor with the mutation threshold; S4. When the voltage mutation factor reaches the mutation threshold, stop the listening task, perform boundary optimization based on the transient voltage sequence and thermal breakdown critical value, and obtain the limit flash time window. S5. Based on the closing of the control switch under the limit flash time window, the surge energy of the energy storage capacitor is injected into the LED light source, driving the LED light source to irradiate the hot colored plastic layer laid on the contact surface, and the detection module receives the red and blue reflected voltages reflected by the hot colored plastic layer. S6. Introduce the voltage mutation factor into the compensation model to correct the red and blue reflected voltages, calculate the color transfer index, write the color transfer index into the non-volatile memory for latching, and output a conduction signal to the bypass switch to short-circuit the secondary side of the CT power supply unit.

2. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S1, during normal operation, the edge computing module acquires the micro-disturbance tilt sequence of the fuse body's fusible element through a triaxial tilt sensor, and collects the transient voltage sequence of the secondary side energy storage capacitor of the CT power extraction unit through a sampling circuit, including: The hardware communication interface of the triaxial tilt sensor is activated, and the triaxial acceleration components are continuously read according to the set sampling clock cycle. The read triaxial acceleration components are then transformed by a coordinate rotation matrix to extract the micro-perturbation tilt sequence that characterizes the deviation of the vertical axis of the molten material. The high-impedance voltage divider branch in the sampling circuit is turned on to synchronously capture the analog level across the energy storage capacitor. The captured analog level is then discretized and calculated via the true RMS conversion module to generate the transient voltage sequence.

3. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S2, the step of extracting phase space features based on the micro-perturbation tilt angle sequence and generating the mechanical stiffness index includes: Extract the absolute value of the tilt angle difference between adjacent sampling points in the micro-perturbation tilt angle sequence, and calculate the arithmetic mean to obtain the absolute motion trajectory evolution rate; Calculate the root mean square amplitude of the micro-perturbation tilt angle sequence in the current domain to obtain the potential energy fluctuation of the system; The quotient is obtained by dividing the absolute motion trajectory evolution rate by the system potential energy fluctuation, and then multiplying the quotient by the time constant correction benchmark to generate the dimensionless mechanical stiffness index.

4. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S2, the step of outputting fatigue warning information based on the mechanical stiffness index includes: Extract the factory mechanical stiffness reference value pre-stored in the storage unit; The mechanical stiffness index is compared with the factory mechanical stiffness reference value. If the comparison result indicates that the mechanical stiffness index is greater than the factory mechanical stiffness reference value of the first set ratio, the fatigue warning information of dry friction jamming of the rotating shaft is output. If the comparison result indicates that the mechanical stiffness index is less than the factory mechanical stiffness benchmark value of the second set ratio, fatigue warning information indicating softening of the return spring stiffness is output.

5. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S3, the step of performing a differential operation based on the transient voltage sequence to obtain the voltage change factor includes: Extract the transient voltage sequence of two adjacent differential time steps, perform a subtraction operation and divide by the corresponding differential time step to obtain the actual voltage change rate; Extract the preset extreme constant of the charging slope at the safe voltage; Calculate the ratio of the actual voltage change rate to the extreme constant of the safe voltage charging slope, and multiply the ratio by the proportional linkage coefficient to generate the dimensionless voltage mutation factor.

6. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S4, the step of performing boundary optimization based on the transient voltage sequence and the thermal breakdown critical value to obtain the limiting flash time window includes: Based on the transient voltage sequence at the current sampling time and the inherent capacitance value of the energy storage capacitor, a dynamic energy function characterizing the accumulated energy inside the energy storage capacitor is reconstructed. Calculate the energy difference between the thermal breakdown critical value and the dynamic energy function to obtain the remaining safe injection energy; Extract the rated power of the LED light source, calculate the ratio of the remaining safe injected energy to the product of the voltage mutation factor and the rated power, and use the calculated ratio as the limiting flash time window when the energy time derivative is in a negative constant state.

7. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S5, the step of controlling the switch to close based on the limit flash time window, injecting the surge energy of the energy storage capacitor into the LED light source, and driving the LED light source to irradiate the hot-colored plastic layer applied to the contact surface includes: Under the condition that the starting moment of the limit flash time window is reached, the control switch is triggered to switch from the open state to the closed state; The surge energy in the energy storage capacitor, which increases logarithmically, is guided to bypass the voltage regulation and buck circuit and be directly input to the positive terminal of the LED light source; The LED light source is excited by a transient overload current exceeding the rated operating power to emit a transient penetrating flash, which illuminates the hot-colored plastic layer covered by the target dirt.

8. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S6, the voltage mutation factor is introduced into the compensation model to correct the red and blue reflection voltages, and the color transfer index is calculated as follows: Extract the excitation charge diffusion constant, which characterizes the bandgap under overload injection conditions, and calculate the product of the natural logarithm of the voltage mutation factor and the excitation charge diffusion constant to generate a dynamic compensation base. The red reflection voltage is multiplied by the correction coefficient obtained by the superposition constant of the dynamic compensation base to obtain the corrected red reflection component, while the blue reflection voltage is kept as the original component. The dimensionless color transfer index is generated by performing a ratio division operation by subtracting the corrected red reflection component from the blue reflection voltage as the numerator and adding the corrected red reflection component to the blue reflection voltage as the denominator.

9. The intelligent monitoring method for fuses according to claim 1, characterized in that, In S6, writing the color transfer index into the non-volatile memory latch and outputting an on signal to the bypass switch to short-circuit the secondary side of the CT power supply unit includes: Enable the direct memory access channel inside the edge computing module and write the color transfer index into the physical address block specified by the non-volatile memory; In a machine cycle that detects that the write completion flag returned by the direct memory access channel is set, a low cutoff level is output to the control switch to cut off the power supply circuit of the LED light source. The saturated conduction level signal is output synchronously to the control electrode of the bypass switch, and the subsequent surge energy generated by the secondary side of the CT power supply unit is discharged to the ground terminal through the drain-source channel.

10. A fuse intelligent monitoring system, characterized in that, A fuse intelligent monitoring method according to any one of claims 1-9, comprising the following modules: A fuse body, the fuse body comprising contacts and a fusible element, wherein the surface of the contacts is coated with a heat-colored plastic layer; A three-axis tilt sensor is fixed on the molten metal carrier to acquire the micro-disturbance tilt angle sequence of the molten metal carrier. A sampling circuit, which is connected to the energy storage capacitor, is used to acquire the transient voltage sequence of the energy storage capacitor; The CT power supply unit has the energy storage capacitor connected in parallel on its secondary side. The positive terminal of the energy storage capacitor is connected to the positive terminal of the LED light source through a first branch series control switch and to the drain of the bypass switch through a second branch. The source of the bypass switch is grounded. A detection module is positioned against the thermochromic plastic layer to receive the red and blue reflected voltages from the thermochromic plastic layer. Non-volatile memory, connected to the edge computing module, is used to latch color transfer indices; The edge computing module is electrically connected to the triaxial tilt sensor, the sampling loop, the control switch, the detection module, the non-volatile memory, and the bypass switch. The edge computing module internally includes: The background listening unit is used to schedule the micro-disturbance tilt sequence through the triaxial tilt sensor during normal operation, and synchronously transmit the transient voltage sequence through the sampling loop. The mechanical early warning unit is used to extract phase space features based on the micro-perturbation tilt angle sequence, generate a mechanical stiffness index, and output fatigue early warning information based on the mechanical stiffness index. A transient differential unit is used to perform differential operations based on the transient voltage sequence to obtain a voltage mutation factor, and continuously compare the voltage mutation factor with a mutation threshold. A boundary optimization unit is used to stop the listening task when the voltage mutation factor reaches the mutation threshold, and to perform optimization calculations based on the transient voltage sequence and the thermal breakdown critical value to obtain the limit flash time window. An energy hijacking drive unit is used to control the closing of the control switch based on the extreme flash time window, inject the surge energy of the energy storage capacitor into the LED light source to emit penetrating flashes to irradiate the hot-colored plastic layer, and simultaneously read the red and blue reflected voltages collected by the detection module; The self-healing latch unit is used to introduce the voltage mutation factor into the compensation model to correct the red and blue reflected voltages in order to calculate the color transfer index, write the color transfer index into the non-volatile memory for latching, and output a conduction signal to the bypass switch to physically short-circuit the secondary side of the CT power supply unit.

Images