Online detection device for melting body of fuse based on time-sharing injection current method

Abstract

The application relates to the technical field of state monitoring, and discloses a fuse body on-off online detection device based on a time-sharing injection current method, which comprises the following: a main circuit working condition monitoring module which outputs a time-sharing trigger instruction when a main circuit current is in a main circuit zero-crossing judgment interval; a multi-frequency excitation injection module which receives an instruction and injects a multi-frequency composite test alternating current, and multiplexes the multi-frequency composite test alternating current into an electrical impedance test source and an excitation source of acoustic elastic waves; a multi-physical field synchronous acquisition module which acquires a response voltage and an acoustic signal; a characteristic decoupling extraction module which analyzes a multi-frequency band electrical impedance characteristic vector and extracts acoustic emission energy characteristics with the time-sharing trigger instruction boundary as a timing reference; and a fault prediction and health management module which uses a first frequency band low-frequency real part impedance to determine on-off, carries out Markov distance synchronous abnormal diagnosis and characteristic weight adjustment control, and fuses and calculates a health degree index. The application cooperates physical avoidance and algorithm timing, constructs an electrical-acoustic homologous alignment mechanism, and blocks parameter jump caused by pulse interference.

Description

Technical Field

[0001] This invention relates to the field of condition monitoring technology, specifically to an online detection device for the continuity of fuse elements based on the time-division current injection method. Background Technology

[0002] Online monitoring of fuse on / off status during industrial main circuit operation typically relies on real-time extraction of voltage drop across the fuse terminals or loop impedance. Existing online monitoring devices often directly perform voltage capture or test signal injection operations while the main circuit load current flows through the fuse body. When the main circuit is in operation, alternating electromagnetic fields and electromagnetic pulses generated by the start-up and shutdown of line loads directly radiate to the physical data acquisition channel. When processing weak response signals, the data sensing hardware nodes cannot isolate background environmental noise crosstalk, and the electrical parameters extracted by the test circuit are mixed with waveform glitches and distortion components, causing baseline drift in the front-end acquisition characteristics. Hardware filtering circuits alone cannot eliminate background interference signals overlapping with the test frequency band, causing the data sequence acquired by the terminal microprocessor to deviate from the true physical state of the fuse.

[0003] Existing fuse condition assessment logic primarily relies on single-dimensional purely electrical characteristics to construct the judgment model. The degradation process of the conductive medium within the fuse structure is accompanied by gradual changes in microscopic volume expansion and mechanical stress; simply recording electrical impedance parameters cannot comprehensively map the multi-physics fatigue degree of the fuse material. In industrial field testing environments, frequency-domain coupling phenomena exist between spatially distributed capacitance and parasitic inductance of the leads, leading to aliasing of real and imaginary impedance characteristic data. When the electrical acquisition channel encounters transient electromagnetic shocks causing characteristic jumps, existing assessment systems lack cross-validation mechanisms across physical field data and parameter weight adjustment algorithms. The model directly reads the abrupt change in value and outputs a health index, causing a sharp drop in the calculated value, prompting the monitoring host to issue false alarm warnings, hindering the smooth quantitative tracking of the fuse degradation trend. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an online fuse element continuity detection device based on the time-division current injection method. This device solves the problems of existing online fuse detection processes being susceptible to signal distortion due to strong electromagnetic noise interference from the main circuit, difficulty in accurately determining fuse continuity under high-frequency parasitic parameter coupling environments, reliance on single pure electrical characteristics for evaluation which is highly susceptible to transient impacts leading to false alarms in health status warnings, and lack of real-time conversion of abnormal detection results into system adaptive fault-tolerant control trigger sources.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an online fuse element continuity detection device based on the time-division current injection method, comprising: The main circuit condition monitoring module is used to acquire the main circuit current flowing through the fuse under test in real time, and output a time-division trigger command when the main circuit current is detected to be within the preset main circuit zero-crossing judgment interval. The main circuit zero-crossing judgment interval is the current amplitude judgment standard preset by the system. The multi-frequency excitation injection module is connected to the main circuit condition monitoring module and is used to inject multi-frequency composite test AC current into the fuse under test after receiving the time-division trigger command. The multi-physics synchronous acquisition module is used to simultaneously acquire the electrical response voltage signal of the fuse under test and the acoustic elastic wave signal induced by the multi-frequency composite test AC current during the injection of the multi-frequency composite test AC current. The feature decoupling extraction module is connected to the multi-physics synchronous acquisition module. It is used to analyze the electrical response voltage signal and the acoustic elastic wave signal, extract multi-band electrical impedance feature vectors, and extract acoustic emission energy features using the available detection window boundary established by the time-division trigger command as the timing reference for the integration start time and integration window length. The fault prediction and health management module is connected to the feature decoupling extraction module. It is used to determine the on / off state of the fuse based on the multi-band electrical impedance feature vector, and to calculate the health index of the tested fuse by fusing the multi-band electrical impedance feature vector with the acoustic emission energy feature.

[0006] Furthermore, the main circuit condition monitoring module is specifically used to: determine in real time whether the main circuit current is within the main circuit zero-crossing determination interval; when the main circuit current is within the main circuit zero-crossing determination interval, start a duration timer; if the duration of the main circuit current being within the main circuit zero-crossing determination interval is greater than or equal to a preset minimum relaxation time threshold, then determine that the current time period is an available detection window, and output the time-sharing trigger command to trigger detection.

[0007] Further, the multi-frequency excitation injection module is specifically used to: synthesize the multi-frequency composite test AC current containing a first frequency band component and a second frequency band component, wherein the first frequency band component is configured as a low-frequency test frequency band for weakening parasitic parameter interference, the second frequency band component is configured as a high-frequency excitation frequency band for inducing the acoustic elastic wave signal, the first frequency band is lower than the second frequency band, and the initial phase of each frequency band component is configured by a phase modulation algorithm so that the peak factor of the synthesized current sequence is less than a preset peak factor threshold; inject the synthesized multi-frequency composite test AC current into the fuse under test so as to reuse the multi-frequency composite test AC current as an electrical impedance test excitation source and a physical excitation source for generating the acoustic elastic wave signal.

[0008] Furthermore, the multi-physics synchronous acquisition module includes: a differential amplifier circuit, connected across the two ends of the fuse under test, for capturing the electrical response voltage signal; and a piezoelectric ceramic pickup, arranged outside the fuse under test, for synchronously capturing the acoustic elastic wave signal generated by the fuse under test due to electromagnetic force and thermal expansion.

[0009] Furthermore, the feature decoupling extraction module is specifically used to: perform discrete Fourier transform on the injected multi-frequency composite test AC current and the acquired electrical response voltage signal to obtain the complex current vector and complex voltage vector corresponding to each preset frequency point in the first frequency band component and the second frequency band component; calculate the complex impedance at each preset frequency point based on the complex current vector and the complex voltage vector, and combine the obtained real part impedance and imaginary part impedance in the frequency domain at each frequency point to construct the multi-frequency band electrical impedance feature vector.

[0010] Furthermore, the feature decoupling extraction module is also used to: filter the acoustic elastic wave signal using a digital bandpass filter to remove low-frequency mechanical environmental vibration noise and retain the signal frequency band covering the frequency range of the second frequency band component; determine the integration start time and integration duration based on the available detection window boundary established by the time-division triggering command; and perform square integration calculation on the filtered signal within a specified time period starting from the integration start time and with a length equal to the integration duration to obtain the acoustic emission energy characteristics used to quantify the degree of high-frequency mechanical attenuation.

[0011] Furthermore, when the fault prediction and health management module performs fuse on / off determination, it is specifically used to: extract the low-frequency real part impedance at a preset frequency point corresponding to the first frequency band component in the multi-frequency band electrical impedance feature vector; compare the low-frequency real part impedance with the system preset physical open circuit threshold; if the low-frequency real part impedance is greater than the physical open circuit threshold, then determine and output the status signal that the tested fuse has been disconnected.

[0012] Furthermore, when it is determined that the tested fuse has not tripped, the fault prediction and health management module calculates the health index of the tested fuse for the current period. The core calculation formula is as follows:

[0013] in: Health index for the current period. : The historical health index obtained from the previous testing period. The electrical impedance degradation rate function value is calculated by substituting the multi-band electrical impedance characteristic vector of the current cycle into a preset electrical impedance degradation model, which is obtained by fitting empirical data of fuse full life cycle aging test under laboratory environment. The acoustic energy degradation rate function value is calculated by substituting the acoustic emission energy characteristics into a preset acoustic energy degradation model, which is obtained by fitting historical experience data. The first weighting coefficient represents the weight of electrical characteristics in the system health assessment. The second weighting coefficient represents the weight of acoustic emission characteristics in the system health assessment. and The initial baseline values ​​were calibrated offline based on a multiphysics historical experience dataset, satisfying the following requirements. . The time interval step between two adjacent available detection windows.

[0014] Furthermore, to address the evaluation distortion caused by complex electromagnetic interference in industrial settings, the fault prediction and health management module introduces a cross-validation and dynamic weight adjustment mechanism based on Mahalanobis distance before calculating the health index. Specifically, it calculates the Mahalanobis distance between the current period's multi-band electrical impedance characteristic vector and the historical average electrical impedance characteristic vector. To configure the Mahalanobis distance synchronization as a detector for anomaly diagnosis and a trigger source for a robust, fault-tolerant controller:

[0015] in: : The multi-band electrical impedance feature vector extracted in the current cycle. The historical electrical impedance characteristic mean vector is obtained through statistical calculation based on the historical normal operation dataset of the tested object. The covariance matrix, used to characterize the correlation between various dimensions of multi-band impedance features, is calculated based on historical normal experience datasets. It is the inverse matrix.

[0016] When the calculated Mahalanobis distance When the distance exceeds the system's preset threshold, the system determines that there is an anomaly in feature acquisition (such as sudden strong electromagnetic interference causing electrical signal distortion). Simultaneously, the system uses the anomaly detection result to drive the dynamic adjustment of the weights. At this time, the device dynamically reduces the first weight coefficient. And increase the second weighting coefficient accordingly. By reducing the proportion of abnormal electrical characteristics and increasing the proportion of acoustic characteristics with strong electromagnetic interference resistance, the robustness of the health index assessment is ensured.

[0017] Furthermore, the fault prediction and health management module compares the calculated health index for the current period with the system's preset health warning threshold; if the health index is lower than the health warning threshold, a degradation warning signal is generated, and the health index and the degradation warning signal are output to the host computer through the communication interface.

[0018] This invention breaks through the technical bottleneck of traditional online monitoring relying on pure electrical signals, which makes it susceptible to interference, by using a zero-intersection time-division injection mechanism and an electroacoustic dual-physics field multiplexing excitation architecture. Furthermore, by combining Mahalanobis distance quantization anomaly and multi-feature dynamic weight fusion algorithm, it achieves absolute determination of the fuse's on / off state and high-precision, high-reliability online quantitative assessment of degradation trends.

[0019] This invention provides an online fuse element continuity detection device based on the time-division current injection method. It has the following advantages: 1. The main circuit condition monitoring module of the present invention outputs a time-division trigger command in the zero-crossing determination interval of the main circuit, synchronously triggers the injection of test current and sets the integral boundary for acoustic emission energy feature extraction. A single command works together to connect the physical isolation mechanism of electromagnetic noise in the main circuit with the time axis of terminal feature integration calculation, thereby cutting off the crosstalk of background interference to the start time of feature acquisition from the source.

[0020] 2. The multi-frequency composite test AC current of the present invention serves as an electrical measurement source for extracting impedance characteristics in parallel, and a physical excitation source for exciting acoustic elastic waves based on the Joule heating and electromagnetic force effects. A single electrical port input replaces the external acoustic probe, eliminating spatial misalignment and trigger delay of the separation excitation, and forcibly establishing a spatiotemporal alignment mechanism of response characteristics across physical fields.

[0021] 3. This invention retrieves the real part impedance of the low frequency of the first frequency band to eliminate reactive reactance coupling and performs on / off judgment. It also calculates the Mahalanobis distance of the characteristic vector of electrical impedance of multiple frequency bands. When the Mahalanobis distance value exceeds the limit and triggers the abnormal diagnosis flag, it synchronously drives the adjustment of the weight allocation ratio of the dual-dimensional feature fusion. The single judgment logic is combined to perform abnormal monitoring and fault-tolerant control, blocking parameter jumps caused by transient electromagnetic shocks. Attached Figure Description

[0022] Figure 1 This is a system control principle and internal functional module logic block diagram of the present invention; Figure 2 This is a schematic diagram of the overall structure of the present invention; Figure 3 This is a flowchart of the overall control process for online detection based on the time-division current injection method of the present invention; Figure 4 This is a schematic diagram of the timing waveforms of the main circuit zero-cross triggering and time-division detection of the present invention; Figure 5This is a schematic diagram of the multidimensional physical field feature decoupling and state evaluation algorithm of the present invention; Figure 6 This is a schematic diagram comparing the health index degradation and Mahalanobis distance interference resistance of the present invention.

[0023] Explanation of reference numerals in the attached diagram: 1. Fuse body; 2. First conductive terminal; 3. Second conductive terminal; 4. Main circuit busbar; 5. Main current transformer; 6. First detection probe; 7. Second detection probe; 8. Excitation injection wire; 9. Measurement and control main unit box; 101. Main circuit condition monitoring module; 102. Multi-frequency excitation injection module; 103. Multi-physics field synchronous acquisition module; 104. Feature decoupling extraction module; 105. Fault prediction and health management module; 1031. Differential amplifier circuit; 32. Piezoelectric ceramic pickup. Detailed Implementation

[0024] 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.

[0025] Please see the appendix Figure 1 To be continued Figure 6 This invention provides an online detection device for the continuity of fuse elements based on the time-division current injection method. This device relies on a preset hardware entity combined with specific functional modules to complete the extraction and state assessment of multi-physics field signals.

[0026] The online fuse element continuity detection process based on the time-division current injection method may include the following steps: Real-time monitoring of the main circuit current condition is used to determine the zero-crossing judgment range and trigger window of the main circuit. In response to a trigger command, inject multi-frequency composite test AC current; Simultaneously acquire electrical response voltage signals and acoustic elastic wave signals; Decouple and extract multi-band electrical impedance feature vectors and acoustic emission energy features; The system integrates multi-dimensional dynamic features to determine the on / off state of the melt and calculate the health index.

[0027] The testing device mainly includes a main circuit bus 4, a fuse under test connected in series in the main circuit bus 4, and a measurement and control host box 9. The fuse under test has a fuse body 1 with a cylindrical insulating shell. The fuse body 1 has a first conductive terminal 2 and a second conductive terminal 3 respectively at both ends. The main circuit bus 4 is horizontal and connects the terminals at both ends. The measurement and control host box 9 integrates a main circuit condition monitoring module 101, a multi-frequency excitation injection module 102, a multi-physics field synchronous acquisition module 103, a feature decoupling extraction module 104, and a fault prediction and health management module 105.

[0028] The main circuit condition monitoring module 101 acquires the main circuit current sequence flowing through the fuse under test in real time using a ring-shaped main current transformer 5 mounted on one side of the main circuit bus 4. After acquiring the sequence, the main circuit condition monitoring module 101 calculates and determines in real time whether the amplitude of the current in the main circuit is within the system's preset zero-crossing judgment interval, which is a current amplitude judgment standard preset by the system. When the current in the main circuit is within the zero-crossing judgment interval, the main circuit condition monitoring module 101 triggers an internal hardware timer to start a duration timing operation. If the duration of the current in the zero-crossing judgment interval is greater than or equal to the preset minimum relaxation time threshold, the main circuit condition monitoring module 101 determines that the current time period is a usable detection window for stable main circuit conditions, and then outputs a time-division trigger command to the multi-frequency excitation injection module 102 to trigger the active detection process. The aforementioned zero-crossing determination interval and minimum relaxation time threshold of the main circuit were obtained offline by the system based on the load fluctuation characteristics of the main circuit in the industrial field and historical waveform statistical experience data. This mechanism aims to avoid electromagnetic noise interference in the high-load operating range of the main circuit. At the same time, this time-division triggering command serves both as a physical signal level to avoid interference and as a timing reference at the algorithm feature extraction level, bridging the two completely different technical levels of physical interference avoidance and algorithm timing through a single command.

[0029] The multi-frequency excitation injection module 102 maintains a bus connection with the main circuit condition monitoring module 101. Upon receiving a time-division trigger command, the multi-frequency excitation injection module 102 initiates its internal signal synthesis algorithm to synthesize a multi-frequency composite test AC current containing a first frequency band component and a second frequency band component. The first frequency band component is configured as a low-frequency test frequency band to reduce parasitic parameter interference, and the second frequency band component is configured as a high-frequency excitation frequency band to induce the acoustic elastic wave signal, with the first frequency band being lower than the second frequency band. The multi-frequency excitation injection module 102 employs a nonlinear phase modulation algorithm (e.g., the Schroeder phase modulation algorithm or a genetic optimization algorithm) to optimize the initial phase configuration of each frequency band component, ensuring that the peak factor of the synthesized multi-frequency composite test AC current sequence is limited to a range less than a preset peak factor threshold.

[0030] The time-domain synthesis formula for AC current in multi-frequency composite testing is expressed as follows:

[0031] In the formula, Indicates time The multi-frequency composite test AC current output by the etching process; This represents the total number of discrete frequency points contained in the first frequency band component; and These represent the components in the first frequency band. Current amplitude and signal frequency at each frequency point; Indicates the first frequency band component Each frequency point is configured with its initial phase based on the phase modulation algorithm; This represents the total number of discrete frequency points contained in the second frequency band component; and These represent the components in the second frequency band. Current amplitude and signal frequency at each frequency point; Indicates the second frequency band component The initial phase of each frequency point is configured, and the amplitude and frequency parameters of each frequency point are set in advance based on the empirical model of the rated impedance of the fuse.

[0032] The preset peak factor control formula is expressed as follows:

[0033] In the formula, This represents the calculated peak factor value; This represents the complete common period of the synthesized test AC current sequence; The preset peak factor threshold is determined by the system based on the full-scale current output limit of the signal generation circuit hardware.

[0034] After completing the current sequence synthesis operation, the multi-frequency excitation injection module 102 is connected to the first conductive terminal 2 and the second conductive terminal 3 via two parallel excitation injection wires 8, respectively, to inject the multi-frequency composite test AC current into the fuse under test. During the injection detection window, the multi-frequency composite test AC current is multiplexed as both an electrical impedance test excitation source and a physical excitation source. The principle of the physical excitation source is that when the AC current flows through the fusible body with a specific resistivity, the alternating electromagnetic repulsion and Joule heating cause microscopic volume thermal expansion, generating periodic stress. This stress is transmitted to the periphery, inducing the fuse body 1 to generate acoustic elastic wave signals with corresponding frequency band distribution, realizing the homogeneous coupling multiplexing of the electroacoustic dual physical fields.

[0035] The multi-physics synchronous acquisition module 103 performs synchronous acquisition of electrical response voltage signals and acoustic elastic wave signals during the injection of multi-frequency composite test AC current by the multi-frequency excitation injection module 102. Specifically, the multi-physics synchronous acquisition module 103 includes a differential amplifier circuit 1031 and a piezoelectric ceramic pickup 32. The voltage input terminal of the differential amplifier circuit 1031 forms a measurement circuit across the two ends of the fuse under test via a first detection probe 6 perpendicularly abutting the top of the first conductive terminal 2 and a second detection probe 7 perpendicularly abutting the top of the second conductive terminal 3, to capture the electrical response voltage signal generated under the action of the multi-frequency composite test AC current. The piezoelectric ceramic pickup 32 adopts a square thin-film structure, closely attached to the middle of the insulating shell of the fuse body 1, to synchronously capture the acoustic elastic wave signals transmitted to the surface by the internal structure of the fuse under multiplexing excitation. The main current transformer 5, the first detection probe 6, the second detection probe 7, and the piezoelectric ceramic pickup 32 are all connected to the measurement and control host box 9 via independently shielded coaxial signal cables. The shielded channel isolates external interference to ensure signal transmission and enables synchronous acquisition of the electrical response voltage signal and the acoustic elastic wave signal under the same reference clock domain, providing the original data source for the data decoupling and extraction process.

[0036] In this embodiment, after receiving the electrical response voltage signal and acoustic elastic wave signal transmitted by the synchronous acquisition module, the feature decoupling extraction module 104 initiates the analysis and feature quantization process of the multi-dimensional physical field signal. The device independently executes processing algorithms corresponding to the electrical and acoustic signals using the feature decoupling extraction module 104, extracting multi-band electrical impedance feature vectors and acoustic emission energy features respectively, thereby achieving deep decoupling and dimensional alignment of cross-physical field data features.

[0037] The feature decoupling extraction module 104 acquires the discrete sequence of injected multi-frequency composite test AC current output by the multi-frequency excitation injection module 102, and simultaneously acquires the discrete sequence of electrical response voltage signal captured by the differential amplifier circuit 1031. For the above two sets of time-domain signal sequences, the feature decoupling extraction module 104 simultaneously performs a discrete Fourier transform operation to map the fluctuation data in the time-domain dimension to the frequency-domain dimension, thereby acquiring the complex current vector and complex voltage vector corresponding to each preset frequency point in the first and second frequency band components.

[0038] For any preset frequency point, the corresponding discrete Fourier transform formula and its complex vector mathematical expression are as follows:

[0039]

[0040] In the formula, This indicates the configuration of the first frequency band component and the second frequency band component. One preset frequency point; This represents the total number of discrete sampling points within the detection window corresponding to the time-division trigger command; Represents a discrete time series index; This represents the discrete-time sampled value of the injected multi-frequency composite test AC current; This represents the discrete-time sampled value of the synchronously acquired electrical response voltage signal; This represents the preset frequency point. The complex vector of the current at that point; This represents the preset frequency point. The complex voltage vector at that point; It is the imaginary unit.

[0041] After obtaining the complex vector, the feature decoupling extraction module 104 calculates the preset frequency point based on the circuit node calculation rules. The complex impedance at that point. The formula for calculating this impedance is:

[0042] In the formula, Indicates the corresponding preset frequency point Calculate the complex impedance at the location; The frequency domain real part impedance represents the complex impedance result and is used to characterize the pure resistive properties of the conductive medium inside the fuse. The frequency-domain imaginary impedance represents the complex impedance result and is used to reflect the combined reactance response characteristics of the system composed of parasitic inductance and distributed capacitance.

[0043] After completing the calculation of complex impedance data at each preset frequency point, the feature decoupling extraction module 104 sequentially combines the extracted real and imaginary impedances at each frequency point to construct a multi-band electrical impedance feature vector. The matrix is ​​constructed in the following form:

[0044] In the formula, The number of preset frequency points contained in the first and second frequency band components constitutes the basic frequency domain input data for quantifying the degree of degradation of the internal conductive path of the fuse.

[0045] Within the same time period as the electrical feature extraction step, the feature decoupling extraction module 104 performs a time-domain energy analysis process on the acoustic elastic wave signal captured by the piezoelectric ceramic pickup 32. The feature decoupling extraction module 104 calls its internally loaded digital bandpass filter algorithm to perform filtering operations on the input acoustic elastic wave signal sequence. The passband frequency range of this digital bandpass filter matches the frequency range of the second frequency band component in the multi-frequency composite test AC current, thereby eliminating low-frequency mechanical environmental vibration noise and background white noise interference, and thus retaining a high-purity signal band that completely covers the frequency range of the second frequency band component.

[0046] After acquiring the filtered high-frequency acoustic signal sequence, the feature decoupling extraction module 104 obtains the integration start time based on the available detection window boundary established by the time-division triggering command. and integration window length The integral parameter is used as the timing reference to perform a square integral calculation, outputting the acoustic emission energy characteristic used to quantify the degree of high-frequency mechanical attenuation. The formula for calculating the square integral of this acoustic emission energy characteristic is expressed as:

[0047] In the formula, This represents the calculated acoustic emission energy characteristics. This refers to the starting trigger time node of the detection window corresponding to the time-sharing trigger command; This represents the duration of the detection window; This represents the acoustic elastic wave time-domain voltage signal output after processing by a digital bandpass filter.

[0048] In this embodiment, the device relies on the fault prediction and health management module 105 to execute an absolute determination procedure for the fuse's on / off state and a health index calculation algorithm based on multi-dimensional feature fusion. The fault prediction and health management module 105 extracts the low-frequency real impedance at a preset frequency point within the first frequency band component of the multi-band electrical impedance feature vector and compares it with the physical open-circuit threshold stored in the system. The fault prediction and health management module 105 selects this low-frequency real impedance as the determination criterion, taking into account both measurement accuracy and anti-interference, and uses this low-frequency real impedance index to avoid frequency domain coupling interference caused by parasitic inductance and distributed capacitance. If the low-frequency real impedance is greater than the physical open-circuit threshold, the fault prediction and health management module 105 determines that the fuse is in the open state, and then outputs a status signal and terminates the health index calculation process.

[0049] If the low-frequency real impedance is less than or equal to the physical open-circuit threshold, the fault prediction and health management module 105 determines that the fuse is in a connected state and performs feature cross-validation and dynamic weight adjustment mechanism before starting the health index calculation program. The fault prediction and health management module 105 extracts the multi-band electrical impedance feature vector of the current cycle, calculates the Mahalanobis distance between the feature vector and the historical electrical impedance feature mean vector, and synchronously configures the Mahalanobis distance as a detector for anomaly diagnosis and a trigger source for a robust fault-tolerant controller.

[0050] The formula for calculating the Mahalanobis distance is as follows:

[0051] In the formula, This represents the calculated Mahalanobis distance value; The multi-band electrical impedance eigenvector matrix represents the current cycle. This represents the mean vector of historical electrical impedance characteristics. The system generates this mean vector statistically based on a dataset obtained from historical normal data. The covariance matrix represents the correlation distribution between data items of various dimensions of impedance characteristics in multi-band systems. The system also relies on a dataset obtained from historical normal data to establish this covariance matrix. It is the inverse of the covariance matrix; This represents the matrix transpose operator.

[0052] The fault prediction and health management module 105 compares the calculated Mahalanobis distance value with the distance threshold set by the system. When the Mahalanobis distance value exceeds the set distance threshold, the fault prediction and health management module 105, while determining that there is an abnormal feature acquisition condition in the current field environment (such as sudden strong electromagnetic interference), does not simply make a logical judgment, but synchronously and in real time uses the abnormal detection result to drive the dynamic adjustment of subsequent weights; in response to the controller trigger source, the fault prediction and health management module 105 reduces the first weight coefficient according to the step size value given by the system, and at the same time increases the second weight coefficient accordingly. The first weight coefficient corresponds to the fusion ratio of the multi-band electrical impedance characteristic vector in the evaluation model; the second weight coefficient corresponds to the fusion ratio of the acoustic emission energy characteristic in the evaluation model; the system uses the above weight adjustment steps to suppress parameter value fluctuations caused by electromagnetic interference.

[0053] After completing the feature acquisition status verification and weight coefficient adjustment steps, the fault prediction and health management module 105 starts the health index calculation model. The fault prediction and health management module 105 retrieves the historical health index of the previous detection cycle recorded in the memory, substitutes the multi-band electrical impedance feature vector of the current cycle into the set electrical impedance degradation rate function to obtain the electrical feature degradation amount; and substitutes the acoustic emission energy feature into the set acoustic energy degradation rate function to obtain the acoustic feature degradation amount.

[0054] The formula for calculating the current period health index is expressed as follows:

[0055] In the formula, This represents the calculated health index for the current period. This represents the historical health index value retrieved by the system from the previous detection period; This represents the value of degradation calculated based on the electrical impedance degradation rate function. This represents the value of degradation calculated based on the acoustic energy degradation rate function. This represents the adjusted first weighting coefficient; This represents the adjusted second weighting coefficient; This represents the time interval step size between the detection windows corresponding to two adjacent time-division trigger commands. Researchers generated the aforementioned electrical impedance degradation rate function and acoustic energy degradation rate function by fitting data from laboratory physical lifetime decay tests. For example, the electrical impedance degradation rate function can be obtained by fitting an exponential decay function based on the Arrhenius model, and the acoustic energy degradation rate function can be obtained by fitting a Weibull distribution decay model, thereby ensuring the determinism and reproducibility of the characteristic degradation calculation.

[0056] After calculating the health index value for the current period, the fault prediction and health management module 105 extracts the health warning threshold configured by the system and determines the relationship between the health index value and the health warning threshold. If the health index value is lower than the health warning threshold, the internal logic operation unit of the fault prediction and health management module 105 generates a degradation warning signal. Subsequently, the fault prediction and health management module 105 retrieves communication interface resources, specifically configured as an industrial Ethernet interface or a controller area network communication bus interface, performs data packet encoding on the current period's health index value and the generated degradation warning signal, and outputs them to the host computer device along an isolated anti-interference communication link for maintenance personnel to perform scheduling and troubleshooting tasks.

[0057] In this embodiment, the target fuse is connected in series in an industrial-grade main circuit with rated operating parameters of 380 volts and 100 amps. The main circuit condition monitoring module 101 continuously acquires the main circuit current sequence flowing through the target fuse; the system sets the absolute value threshold of the current in the main circuit zero-crossing judgment interval to one amp and configures the minimum relaxation time threshold to two milliseconds. When the absolute value of the main circuit current drops to within one amp and the duration reaches two milliseconds, the system determines that the main circuit zero-crossing judgment interval is met, and then outputs a time-division trigger command to open the available detection window for the current time period.

[0058] After receiving a time-division trigger command, the multi-frequency excitation injection module 102 injects an internally synthesized multi-frequency composite test AC current into the target fuse. The system configures the center test frequency of the first frequency band component to be 200 Hz and the center test frequency of the second frequency band component to be 30 kHz. The system uses a differential amplifier circuit 1031 and a piezoelectric ceramic pickup 32 to synchronously acquire the electrical response voltage signal and the acoustic elastic wave signal within the detection window. The feature decoupling extraction module 104 performs discrete Fourier transform and digital bandpass filtering on the above signal sequence to extract the low-frequency real impedance value at the 200 Hz frequency point, which is the low-frequency test frequency band, and the acoustic emission energy characteristic value within the 30 kHz frequency band.

[0059] The fault prediction and health management module 105 retrieves the low-frequency real impedance value at a 200 Hz frequency point and compares it with a preset physical open-circuit threshold. If the low-frequency real impedance value is at the normal milliohm level and less than the physical open-circuit threshold, the fault prediction and health management module 105 calculates the Mahalanobis distance between the multi-band electrical impedance characteristic vector and the historical electrical impedance characteristic mean vector. The covariance matrix in the initial detection phase is set to exhibit a diagonal distribution, and the Mahalanobis distance value calculated by the system for the current period is 1.5. The system sets the Mahalanobis distance threshold to 3.0. Since 1.5 does not exceed this threshold, the system determines that the current feature acquisition environment is not subject to high-frequency electromagnetic interference. At this time, the system maintains the default setting of the first weighting coefficient at 0.6 and the default setting of the second weighting coefficient at 0.4.

[0060] The system retrieves the historical health index from the previous detection cycle. Setting the initial index value to 100%, the fault prediction and health management module 105 substitutes the current cycle's feature values ​​into the degradation model, determining that the current electrical feature degradation is 0.05% and the acoustic feature degradation is 0.02%. Combining this with the set time interval step size of one between two adjacent detection windows, the system uses the health calculation formula to derive the current cycle's health index value:

[0061] The above calculation results indicate that the target fuse is in a healthy operating range.

[0062] As the operating cycle continues to increase, a sudden strong electromagnetic pulse in the main circuit causes distortion in the electrical acquisition channel. The system extracts abnormal electrical feature data and calculates a new Mahalanobis distance value of 4.8. This value exceeds the set three-point zero distance threshold, and the fault prediction and health management module 105 determines that there is an abnormal feature acquisition condition. At this time, the Mahalanobis distance, as the controller trigger source, immediately takes effect. The system uses this abnormal result to directly drive dynamic adjustment. Based on the given step size, the system lowers the first weight coefficient to 0.2, while simultaneously raising the second weight coefficient corresponding to the acoustic emission energy feature with strong anti-electromagnetic interference capability to 0.8. The system retrieves the historical health index, which had previously dropped to 80.5%, and recalculates that the electrical feature degradation in the current interference cycle shows a jump value of 1.5%, while the acoustic feature degradation remains stable at 0.1%.

[0063] Using the dynamically adjusted weight parameters, the system again uses the health calculation formula to obtain the health index value within the interference period:

[0064] Under this calculation mechanism, the system suppresses the sharp drop in evaluation results caused by abrupt changes in electrical characteristics, ensuring the stability of quantitative data. The system compares the calculated health index value with a health warning threshold set at 80%. As data accumulates in the detection window, when the health index value continues to decay and falls below 80%, the fault prediction and health management module 105 generates a degradation warning signal. The system sends a data packet containing this warning signal and the health index to the host computer via a communication link, completing the task of online monitoring of non-destructive conditions in the industrial environment.

[0065] In this embodiment, the measurement and control host chassis 9 integrates a microprocessor and a computer-readable storage medium connected to the microprocessor via a data communication bus. The microprocessor employs a heterogeneous computing structure combining a digital signal processor (DSP) and a field-programmable gate array (FPGA). The FPGA interfaces with the differential amplifier circuit 1031 and the piezoelectric ceramic pickup 32 in the multi-physics synchronous acquisition module 103, executing time-domain signal acquisition and analog-to-digital conversion timing control logic. The DSP receives the data sequence packaged and output by the FPGA and performs mathematical operations such as discrete Fourier transform, multi-band impedance characteristic decoupling, and Mahalanobis distance verification.

[0066] The computer-readable storage medium uses a non-volatile read-only memory or a flash memory chip. The computer-readable storage medium internally stores computer program code; this computer program code contains an instruction set corresponding to the operational rules of each functional module. When the microprocessor reads and executes the instruction set stored internally in the computer-readable storage medium, it sequentially triggers core algorithm steps such as main circuit current relaxation window timing, multi-frequency composite test AC current waveform matrix synthesis, construction of frequency domain real and imaginary impedance vectors, bandpass filtering and square integral calculation of acoustic elastic wave signals, and multi-dimensional dynamic feature weight adjustment and health index degradation model calculation.

[0067] Based on the instruction execution cycle indicators and device power consumption limits of the terminal industrial scenario, the R&D designers programmed the operation rules of the main circuit condition monitoring module 101, multi-frequency excitation injection module 102, feature decoupling extraction module 104, and fault prediction and health management module 105 into the hardware logic gate circuits inside the microprocessor, or compiled them into system software executable code. The system software executable code, together with the computing system composed of a general-purpose processing chip, is functionally equivalent to an execution entity composed of an independent application-specific integrated circuit when running a multi-physics information fusion judgment program.

[0068] The device is equipped with a communication interface for connecting to an external transmission network. This interface connects to an industrial Ethernet protocol network or a controller area network (CLAN) communication bus. After the microprocessor generates the status assessment result, it encapsulates and outputs the health index value with a synchronization timestamp and the degradation warning signal frame to the externally associated host computer monitoring node via the communication interface. This forms a closed-loop data structure from bottom-level signal perception to top-level status quantification and early warning, completing the full-cycle online status assessment process for the target fuse.

Claims

1. An online fuse element continuity detection device based on the time-sharing current injection method, characterized in that, include: The main circuit condition monitoring module is used to acquire the main circuit current flowing through the fuse under test in real time, and output a time-division trigger command when the main circuit current meets the preset main circuit zero-crossing judgment interval. The main circuit zero-crossing judgment interval is the current amplitude judgment standard preset by the system. The multi-frequency excitation injection module is connected to the main circuit condition monitoring module and is used to inject multi-frequency composite test AC current into the fuse under test after receiving the time-division trigger command. The multi-physics synchronous acquisition module is used to simultaneously acquire the electrical response voltage signal of the fuse under test and the acoustic elastic wave signal induced by the multi-frequency composite test AC current during the injection of the multi-frequency composite test AC current. The feature decoupling extraction module is connected to the multi-physics synchronous acquisition module. It is used to analyze the electrical response voltage signal and the acoustic elastic wave signal, extract multi-band electrical impedance feature vectors, and extract acoustic emission energy features using the available detection window boundary established by the time-division trigger command as the timing reference for the integration start time and integration window length. The fault prediction and health management module is connected to the feature decoupling extraction module. It is used to determine the on / off state of the fuse based on the multi-band electrical impedance feature vector, and to calculate the health index of the tested fuse by fusing the multi-band electrical impedance feature vector with the acoustic emission energy feature.

2. The online fuse element continuity detection device based on time-division current injection method according to claim 1, characterized in that, The main circuit operating condition monitoring module is specifically used for: Real-time determination of whether the main circuit current is within the preset main circuit zero-crossing determination range; When the main circuit current is within the main circuit zero-crossing determination range, the duration timer is started. If the duration of the main circuit current being within the zero-crossing determination interval of the main circuit is greater than or equal to the preset minimum relaxation time threshold, then the current time period is determined to be an available detection window, and the time-sharing trigger command is output to trigger detection.

3. The online fuse element continuity detection device based on time-division current injection method according to claim 1, characterized in that, The multi-frequency excitation injection module is specifically used for: The multi-frequency composite test AC current is synthesized, which includes a first frequency band component and a second frequency band component. The first frequency band component is configured as a low-frequency test frequency band to reduce parasitic parameter interference, and the second frequency band component is configured as a high-frequency excitation frequency band to induce the acoustic elastic wave signal. The first frequency band is lower than the second frequency band, and the initial phase of each frequency band component is configured by a phase modulation algorithm so that the peak factor of the synthesized current sequence is less than a preset peak factor threshold. The synthesized multi-frequency composite test AC current is injected into the fuse under test so that the multi-frequency composite test AC current can be reused as an electrical impedance test excitation source and a physical excitation source for generating the acoustic elastic wave signal.

4. The online fuse element continuity detection device based on time-division current injection method according to claim 1, characterized in that, The multiphysics field synchronous acquisition module includes: A differential amplifier circuit is connected across the two ends of the fuse under test to capture the electrical response voltage signal; A piezoelectric ceramic pickup is arranged on the outside of the fuse under test to synchronously capture the acoustic elastic wave signal generated by the fuse under test due to electromagnetic force and thermal expansion.

5. The online fuse element continuity detection device based on time-division current injection method according to claim 3, characterized in that, The feature decoupling extraction module is specifically used for: Perform a discrete Fourier transform on the injected multi-frequency composite test AC current and the collected electrical response voltage signal to obtain the current complex vector and voltage complex vector corresponding to each preset frequency point in the first frequency band component and the second frequency band component. Based on the complex current vector and the complex voltage vector, the complex impedance at each preset frequency point is calculated, and the real part impedance and the imaginary part impedance at each frequency point are combined to construct the multi-band electrical impedance characteristic vector.

6. The online fuse element continuity detection device based on time-division current injection method according to claim 3, characterized in that, The feature decoupling extraction module is also used for: The acoustic elastic wave signal is filtered using a digital bandpass filter to remove low-frequency mechanical environmental vibration noise and retain the signal frequency band covering the frequency range of the second frequency band component. Based on the available detection window boundaries established by the time-division triggering command, the integration start time and integration duration are determined; The filtered signal is subjected to square integration within a specified time period, starting from the integration start time and lasting for the duration of the integration, to obtain acoustic emission energy characteristics used to quantify the degree of high-frequency mechanical attenuation.

7. The online fuse element continuity detection device based on time-division current injection method according to claim 5, characterized in that, The fault prediction and health management module, when determining the on / off state of the fuse, is specifically used for: Extract the low-frequency real part impedance at a preset frequency point corresponding to the first frequency band component from the multi-band electrical impedance feature vector. The low-frequency real impedance is compared with the system's preset physical open-circuit threshold. If the low-frequency real impedance is greater than the physical open-circuit threshold, the test fuse is determined to be open and a status signal is output.

8. The online fuse element continuity detection device based on time-division current injection method according to claim 1, characterized in that, When calculating the health index, the fault prediction and health management module is specifically used for: When it is determined that the tested fuse has not tripped, the historical health index of the previous testing cycle is obtained; Substitute the multi-band electrical impedance characteristic vector of the current period into a preset electrical impedance degradation rate function, and substitute the acoustic emission energy characteristic into a preset acoustic energy degradation rate function. The health index for the current period is calculated by subtracting the feature degradation amount from the historical health index of the previous detection period; wherein the feature degradation amount is: the product of the electrical impedance degradation rate function and the first weighting coefficient, and the product of the acoustic energy degradation rate function and the second weighting coefficient, and the product of the sum of the two and the time interval step between two adjacent detection windows.

9. The online fuse element continuity detection device based on time-division current injection method according to claim 8, characterized in that, Before calculating the health index, the fault prediction and health management module is also used for: The Mahalanobis distance between the multi-band electrical impedance characteristic vector of the current period and the historical electrical impedance characteristic mean vector obtained based on historical normal data is calculated, so that the Mahalanobis distance is synchronously configured as a detector for anomaly diagnosis and a trigger source for a robust fault-tolerant controller. When the Mahalanobis distance exceeds a preset distance threshold, while determining that there is an anomaly in feature acquisition, the anomaly detection result is used to drive the dynamic adjustment of the weights, reducing the first weight coefficient and correspondingly increasing the second weight coefficient.

10. The online fuse element continuity detection device based on time-division current injection method according to claim 1, characterized in that, The fault prediction and health management module is also used for: The calculated health index for the current period is compared with the system's preset health warning threshold. If the health index is lower than the health warning threshold, a degradation warning signal is generated, and the health index and the degradation warning signal are output to the host computer through the communication interface.

Images