A fault early warning method for abnormal identification of distribution box

Abstract

The application discloses a kind of power distribution box abnormality-oriented fault early warning methods, it is related to power distribution equipment state monitoring and fault early warning field, including the following steps: obtaining the action transient response data of switch device in multiple tripping, closing, attraction and release process in power distribution box, and executing pre-processing;Action feature unit set is constructed;Transient fingerprint vector is constructed, and switch device historical transient fingerprint sequence is generated;Cross-cycle comparison is carried out, and action memory attenuation sequence is formed;Action memory attenuation sequence is input into switch device life evolution determination process, and life evolution instruction sequence is generated;Determine latent fault early warning object, early warning trigger section, risk accumulation grade and fault type indication result;Determine the fault early warning result in power distribution box.The application uses action transient analysis method, realizes power distribution box latent fault early warning, with the advantages of early warning early, identification accurate.

Description

Technical Field

[0001] This invention relates to the field of power distribution equipment condition monitoring and fault early warning, and in particular to a fault early warning method for identifying anomalies in distribution boxes. Background Technology

[0002] Existing fault monitoring technologies for distribution boxes mostly rely on steady-state operating parameters such as temperature, current, voltage, and leakage current. They typically identify and alarm after a switch fails to operate, malfunctions, contacts stick, burn out, or conduction abnormalities occur. They pay insufficient attention to the transient response of the action during the opening, closing, engagement, and release processes.

[0003] However, in the early stages of life decay, the switching devices in the distribution box often exhibit delayed action, enhanced rebound, intensified arcing, and drift in response waveform. Existing technologies lack the ability to continuously correlate transient responses of multiple actions, compare across cycles, and determine degradation evolution, making it difficult to identify latent faults in a timely manner and provide effective early warnings. Summary of the Invention

[0004] One objective of this invention is to propose a fault early warning method for identifying anomalies in distribution boxes. This invention employs a transient action analysis method to achieve early warning of latent faults in distribution boxes, which has the advantages of early warning and accurate identification.

[0005] A fault early warning method for identifying anomalies in distribution boxes according to an embodiment of the present invention includes the following steps: Acquire transient response data of switching devices in the distribution box during multiple opening, closing, energizing and releasing processes, and perform preprocessing to generate action response sequences; Based on the action response sequence, construct a set of action feature units; The action feature unit set is arranged and combined to construct a transient fingerprint vector, and then arranged in chronological order to generate a historical transient fingerprint sequence of the switching device; Cross-cycle comparison is performed on the historical transient fingerprint sequence of switching devices to form an action memory decay sequence; The action memory decay sequence is input into the life evolution determination process of the switching device to generate the segment life evolution index, and arranged in the time order of the preset continuous action segments to generate the life evolution indicator sequence. Based on the lifetime evolution indication sequence, the latent fault warning object, warning triggering section, risk accumulation level and fault type indication result are determined to form a fault warning judgment sequence. The fault warning result in the distribution box is determined based on the fault warning judgment sequence.

[0006] Optionally, the transient response data includes action command time data, coil on / off current data, contact switching time data, action process current waveform data, rebound vibration waveform data, and arc response waveform data.

[0007] Optionally, the preprocessing includes time synchronization, motion segment extraction, noise removal, baseline correction, and device identification association.

[0008] Optionally, the generation of the action feature unit set specifically includes: Read the action process segments corresponding to each single opening, closing, energizing, and releasing action response sequence; Based on the action command time data and the contact conduction switching time data, the action command issuance time and the contact conduction switching time of each action process segment are extracted, and the action delay time is calculated. Based on the coil current data, determine the coil energization start time, the coil current reaches the stable range, and the coil de-energization decay end time, and calculate the engagement setup time and release decay time. Extract the current peak point, impact duration, and waveform deviation from the current waveform data during the action process; extract the current impact peak value, current impact duration, and waveform distortion. By combining rebound vibration waveform data and contact conduction switching time data, contact rebound events are identified, the number of contact rebounds and the total contact rebound time are counted, and the arc start and end times and arc intensity change segments are extracted based on arc response waveform data, and the arc duration and arc intensity integral value are calculated. Summarize the action features to form a set of action feature units.

[0009] Optionally, the generation of the historical transient fingerprint sequence of the switching device includes: Extract the corresponding action feature unit set according to the switching device identifier, and sort each action feature unit according to the action time sequence; The sub-features in the action feature unit are combined to construct the transient fingerprint vector; Arrange the transient fingerprint vectors corresponding to the same switching device identifier in chronological order of action time to generate a historical transient fingerprint sequence of the switching device.

[0010] Optionally, the generation of the action memory decay sequence specifically includes: Read the historical transient fingerprint sequence of the corresponding switching device, and extract adjacent transient fingerprint vectors along the action time sequence to construct multiple preset continuous action segments; For each preset continuous action segment, extract the adjacent transient fingerprint vectors within the segment, compare the features of each dimension of the adjacent transient fingerprint vectors, and generate fingerprint difference values. In each preset continuous action segment, fingerprint drift value and fingerprint discrete value are generated based on the transient fingerprint vector within the segment. For each preset continuous action segment, a rebound deterioration value is generated based on the number of contact rebounds and the total contact rebound time within the segment. Generate an attraction decay value based on the attraction setup time; Based on the release decay time, a release hysteresis value is generated; Based on the arc duration and arc intensity integral value within the section, an arc wear indication value is generated; The fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value and arc wear indication value corresponding to each preset continuous action segment are arranged in sequence to generate an action memory attenuation sequence.

[0011] Optionally, the generation of the lifetime evolution indicator sequence specifically includes: Read the corresponding action memory decay sequence according to the switch device identifier, and extract the data corresponding to each preset continuous action segment; For each preset continuous action segment, a transient response degradation index is generated based on the fingerprint difference value, fingerprint drift value, and fingerprint discrete value. Based on the rebound deterioration value, the pull-in attenuation value, the release hysteresis value, and the arc wear indication value, a mechanism degradation index is generated; Based on the transient response degradation index and the mechanism degradation index, segment deviation value, segment coordination value and segment unidirectional enhancement value are generated; The segment synergy value and the segment co-directional enhancement value are combined, and the segment deviation value is used for suppression correction to generate the segment lifetime evolution index. The lifetime evolution index of each preset continuous action segment is arranged in the time sequence of the preset continuous action segments to generate a lifetime evolution indicator sequence.

[0012] Optionally, the generation of the fault warning determination sequence specifically includes: Read the corresponding lifetime evolution indicator sequence and extract the segment lifetime evolution index corresponding to each segment according to the time sequence of the preset continuous action segments; Compare the segment lifetime evolution index corresponding to adjacent preset continuous action segments to extract segments with continuously rising index and segments with sudden increases in index. By matching the continuous rise in the index with the sudden increase in the index, the potential fault warning targets and warning triggering sections can be identified. Based on the cumulative value of the segment lifetime evolution index and the cumulative value of the segment lifetime evolution index increment within the warning triggering segment, the risk accumulation level of the corresponding switching device is determined; Based on the warning triggering section, determine the fault type indication result; The latent fault warning object, the warning triggering section, the risk accumulation level, and the fault type indication result are correlated to form a fault warning judgment sequence.

[0013] Optionally, the generation of the fault warning result specifically includes: Read the warning triggering section, risk accumulation level and fault type indication results corresponding to each switching device in the fault warning judgment sequence; Extract the corresponding switching device identifier and generate the warning device identifier; Determine the early warning stage based on the level of risk accumulation; Use the cumulative risk level as the risk level; Based on the fault type indication results and the warning stage, generate maintenance prompt information; By associating the warning device identifier, warning stage, warning triggering section, risk level, and maintenance prompt information with the corresponding switch device identifier, the fault warning result can be obtained.

[0014] Optionally, the early warning stage includes an attention stage, an early warning stage, and an emergency early warning stage.

[0015] The beneficial effects of this invention are: Compared to existing methods that primarily rely on steady-state parameters such as temperature, current, voltage, and leakage current for fault identification, this invention acquires transient response data of switching devices within a distribution box during multiple opening, closing, engagement, and release processes. This data is used to construct a set of action feature units, transient fingerprint vectors, and historical transient fingerprint sequences of the switching devices. This allows for continuous characterization of the dynamic behavior of the switching devices from multiple dimensions, including action delay, engagement establishment, release attenuation, current surge, contact rebound, and arcing. By performing cross-cycle comparisons of historical transient fingerprint sequences, an action memory attenuation sequence is formed. Furthermore, a segment lifetime evolution index and lifetime evolution indicator sequence are generated. This enables the invention not only to identify whether a single action is abnormal but also to identify the degradation accumulation, trend changes, and aggravation processes of the same switching device during continuous actions. This advances the fault identification time from after the occurrence of an obvious fault to the latent degradation stage, significantly improving the lead time and sensitivity of distribution box anomaly identification.

[0016] Furthermore, after obtaining the lifetime evolution indication sequence, this invention can determine the latent fault warning object, warning triggering section, risk accumulation level, and fault type indication result, and form a fault warning judgment sequence. Finally, it outputs a fault warning result containing the warning device identifier, warning stage, warning triggering section, risk level, and maintenance reminder information. Therefore, this invention can not only determine whether there is an abnormality in the switching device, but also distinguish different fault types such as transient response abnormalities, contact rebound abnormalities, pull-in attenuation, release hysteresis, and arc wear, and provide warning stages and maintenance reminder information corresponding to the risk level, enhancing the pertinence and executability of the warning results. Because this invention establishes a complete processing link from transient action acquisition, degradation feature extraction, lifetime evolution judgment to fault warning output, it can avoid misjudging isolated fluctuations as faults, while improving the accuracy of identifying latent lifetime faults and the reliability of warnings, which is beneficial for the early maintenance and operational safety of switching devices in distribution boxes. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a fault early warning method for identifying anomalies in distribution boxes proposed in this invention; Figure 2 This is a section life evolution trend curve of a fault early warning method for identifying anomalies in distribution boxes proposed in this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0019] refer to Figure 1 A fault early warning method for identifying anomalies in distribution boxes includes the following steps: Acquire transient response data of switching devices in the distribution box during multiple opening, closing, energizing and releasing processes, and perform preprocessing to generate action response sequences; Based on the action response sequence, construct a set of action feature units; The action feature unit set is arranged and combined to construct a transient fingerprint vector, and then arranged in chronological order to generate a historical transient fingerprint sequence of the switching device; Cross-cycle comparison is performed on the historical transient fingerprint sequence of switching devices to form an action memory decay sequence; The action memory decay sequence is input into the life evolution determination process of the switching device to generate the segment life evolution index, and arranged in the time order of the preset continuous action segments to generate the life evolution indicator sequence. Based on the lifetime evolution indication sequence, the latent fault warning object, warning triggering section, risk accumulation level and fault type indication result are determined to form a fault warning judgment sequence. The fault warning result in the distribution box is determined based on the fault warning judgment sequence.

[0020] In this embodiment, the transient response data includes action command time data, coil on / off current data, contact switching time data, action process current waveform data, rebound vibration waveform data, and arc response waveform data. The action command time data characterizes the time position at which the switching device receives the opening, closing, attracting, and releasing commands. The coil on / off current data characterizes the current change process of the switching device's drive coil during the energizing and de-energizing phases. The contact switching time data characterizes the time position at which the switching device contacts switch from an open state to a conducting state and from a conducting state to an open state. The action process current waveform data characterizes the continuous change of the action branch current over time during the opening, closing, attracting, and releasing processes of the switching device. The rebound vibration waveform data characterizes the vibration response over time caused by mechanical rebound during the contact and separation processes of the switching device contacts. The arc response waveform data characterizes the change of arc intensity over time during the contact separation and re-contact processes of the switching device contacts.

[0021] In this embodiment, preprocessing includes time synchronization, motion segment extraction, noise removal, baseline correction, and device identification association.

[0022] In this embodiment, the generation of the action feature unit set specifically includes: Read the action process segments corresponding to each single opening, closing, energizing, and releasing action response sequence; Based on the action command time data and the contact conduction switching time data, the action command issuance time and the contact conduction switching time of each action process segment are extracted, and the action delay time is calculated. In practice, the action command time data within each action process segment is taken as the starting time position, and the contact switching time data of the first occurrence within the same segment is read as the switching time position. The time difference between the two is calculated to obtain the action delay time of the action. The action delay time of each action is recorded according to the switching device identifier and the action time sequence. The switch device identifier is a unique number used to distinguish the identities of different switch devices in the distribution box. It is used to map the action command time data, coil on / off current data, contact conduction switching time data, action process current waveform data, rebound vibration waveform data, and arc response waveform data to the same switch device, and to continuously associate the switching device with each opening, closing, energizing, and releasing process. Based on the coil current data, determine the coil energization start time, the coil current reaches the stable range, and the coil de-energization decay end time, and calculate the engagement setup time and release decay time. In specific implementation, the time position in the coil on / off current data where the current starts to rise from 0 is found as the coil energization start time; then the starting time position where the current enters the stable range is found as the coil current reaches the stable range time; the time difference between the two is the engagement setup time; the starting time position where the coil on / off current data continuously decreases after reaching the stable range is found, and the time position where the current returns to 0 is found as the coil de-energization attenuation end time; the time difference between the current starting to decrease and the coil de-energization attenuation end time is the release attenuation time. The stable range is the time interval in the coil on / off current data where the current change amplitude remains within a preset fluctuation threshold and continues for a preset duration. The preset fluctuation threshold is ±3% of the average current of the same type of switching device during the coil stable energization phase in normal operation samples, and the preset duration is 20ms. Extract the current peak point, impact duration, and waveform deviation from the current waveform data during the action process; extract the current impact peak value, current impact duration, and waveform distortion. In specific implementation, the current waveform data of the action process is read within a single action process segment, the sampling point where the current reaches its maximum value is located, and the current peak point and the current impact peak point are output. Based on the average steady-state current corresponding to the normal action sample, the continuous time segment where the current is higher than 120% of the benchmark is extracted, the impact duration segment is output, and the start and end time difference of the impact duration segment is calculated, and the current impact duration is output. Then, the current waveform data of the current action process is aligned and compared with the normal action reference waveform of the same type of switching device, the time segment where the absolute value of the difference continuously exceeds the preset deviation threshold is extracted, the waveform deviation segment is output, and the cumulative deviation in each waveform deviation segment is counted, and the waveform distortion is output. In specific implementation, the preset deviation threshold is three times the average absolute value of the difference between the current waveform data of the operation process and the reference waveform of the normal operation in the normal operation sample of the same type of switching device. If it exceeds the threshold continuously, it means that the absolute value of the difference is greater than the preset deviation threshold in no less than 5 consecutive sampling points. Normal operation samples are transient response data of the same type of switching device when it completes the opening, closing, energizing and releasing actions without failure to operate, maloperation, adhesion and burning. Normal operation reference waveforms are standard operating process current waveforms extracted from normal operation samples after being aligned with the operation time. By combining rebound vibration waveform data and contact conduction switching time data, contact rebound events are identified, the number of contact rebounds and the total contact rebound time are counted, and the arc start and end times and arc intensity change segments are extracted based on arc response waveform data, and the arc duration and arc intensity integral value are calculated. In specific implementation, within a single action process segment, the time position when the contact completes its first conduction switch or first disconnection switch is determined by the contact conduction switching time data, and the initial switching time of the contact is output. Then, the rebound vibration waveform data is read around the initial switching time of the contact, and vibration segments with vibration amplitudes continuously exceeding 3 times the vibration baseline corresponding to the normal action sample and lasting for no less than 3 consecutive sampling points are extracted, and candidate rebound segments are output. Each candidate rebound segment is checked against the contact conduction switching time data, and candidate rebound segments occurring within the preset detection window after the initial switching time of the contact are retained, and contact rebound events are output. The number of contact rebound events is counted, the number of contact rebounds is output, and the duration of each contact rebound event is accumulated, and the total contact rebound duration is output. Among them, 3 times the vibration baseline corresponding to the normal action sample is 3 times the average vibration amplitude of the rebound vibration waveform data in the normal action sample of the same type of switching device. In specific implementation, the preset detection window is 30ms after the initial switching time of the contact, and the sampling interval of the continuous sampling points is set to 1ms in specific implementation. Within the same action process segment, read the arc response waveform data, extract the time position when the arc response signal changes from 0 to non-zero, and output the arc start time. Then extract the time position when the arc response signal returns to 0, and output the arc end time. Based on the time difference between the arc start time and the arc end time, output the arc duration. Continue to divide the arc into continuous sampling segments between the arc start time and the arc end time, extract the arc intensity values ​​of each continuous sampling segment and accumulate them, and output the arc intensity integral value. In practice, the zero value of the current and arc response signals is the range of values ​​in which the absolute value after baseline correction does not exceed a preset zero threshold. The preset zero threshold is 1% of the full scale of the corresponding signal. Action delay duration characterizes the speed of response of switching device to action command; engagement setup duration characterizes the setup process required for the drive mechanism to enter a stable engagement state after the coil is energized; release attenuation duration characterizes the attenuation process experienced from the fading of the drive effect after the coil is de-energized to the completion of release; current surge peak value characterizes the current surge intensity at the moment of action; current surge duration characterizes the duration of the current surge state; waveform distortion characterizes the degree of deviation of the current waveform during the action process from the normal action reference waveform; contact rebound count characterizes the frequency of mechanical rebound after contact switching; total contact rebound duration characterizes the cumulative duration of the contact rebound state; arc duration characterizes the duration of the arc during contact separation or re-contact; arc intensity integral value characterizes the cumulative intensity level during the arc action process. The action features are summarized to form a set of action feature units, which include action delay duration, engagement establishment duration, release attenuation duration, current surge peak value, current surge duration, waveform distortion, number of contact rebounds, total contact rebound duration, arc duration, and arc intensity integral value.

[0023] In this embodiment, the generation of the historical transient fingerprint sequence of the switching device includes: Extract the corresponding action feature unit set according to the switching device identifier, and sort each action feature unit according to the action time sequence; The sub-features in the action feature unit are combined to construct the transient fingerprint vector; The transient fingerprint vector is a feature vector formed by sequentially combining the action delay duration, engagement establishment duration, release attenuation duration, current surge peak value, current surge duration, waveform distortion, number of contact rebounds, total contact rebound duration, arc duration, and arc intensity integral value corresponding to the same action. It is used to characterize the transient action response state of the switching device during the opening, closing, engagement, or release process. In practice, the same set of action feature items are extracted for the corresponding action processes of opening, closing, absorbing, and releasing. These feature items include action delay time, absorption establishment time, release attenuation time, current impact peak value, current impact duration, waveform distortion, number of contact rebounds, total contact rebound duration, arc duration, and arc intensity integral value. For feature items that do not have a physical response in a certain action, the feature item is recorded as 0. Arrange the transient fingerprint vectors corresponding to the same switching device identifier in chronological order of action time to generate a historical transient fingerprint sequence of the switching device.

[0024] In this embodiment, the generation of the action memory decay sequence specifically includes: Read the historical transient fingerprint sequence of the corresponding switching device according to the switching device identifier, and extract adjacent transient fingerprint vectors in the order of action time to construct multiple preset continuous action segments; The preset continuous action segment is an analysis segment consisting of a preset number of transient fingerprint vectors selected consecutively in the action time sequence from the historical transient fingerprint sequence of the same switching device. In specific implementation, the preset continuous action segment takes 5 transient fingerprint vectors arranged consecutively in the action time sequence. For each preset continuous action segment, extract the adjacent transient fingerprint vectors within the segment, compare the features of each dimension of the adjacent transient fingerprint vectors, and generate fingerprint difference values. In specific implementation, for each preset continuous action segment, the absolute values ​​of the differences corresponding to the action delay duration, engagement establishment duration, release attenuation duration, current impact peak value, current impact duration, waveform distortion, number of contact rebounds, total contact rebound duration, arc duration, and arc intensity integral value are normalized respectively. The normalized differences of each dimension are weighted and summed, and then the summation results of each adjacent transient fingerprint vector within the segment are accumulated to generate the fingerprint difference value corresponding to the preset continuous action segment. The weights of the weighted summation are set to 0.08, 0.10, 0.10, 0.08, 0.08, 0.12, 0.10, 0.10, 0.12, and 0.12 respectively. In each preset continuous action segment, fingerprint drift value and fingerprint discrete value are generated based on the transient fingerprint vector within the segment. In specific implementation, for each preset continuous action segment, the absolute values ​​of the differences between the action delay duration, engagement establishment duration, release attenuation duration, current impact peak value, current impact duration, waveform distortion, number of contact rebounds, total contact rebound duration, arc duration, and arc intensity integral value between the last transient fingerprint vector and the first transient fingerprint vector are calculated. After normalizing each absolute value of the difference, a weighted sum is generated to produce the fingerprint drift value corresponding to the preset continuous action segment. The weights of the weighted summation are set to 0.10, 0.12, 0.12, 0.07, 0.08, 0.10, 0.10, 0.10, 0.10, and 0.11, respectively. In practice, for each preset continuous action segment, each feature in each transient fingerprint vector within the segment is extracted, and the segment mean of each feature within the preset continuous action segment is calculated. Then, the absolute value of the deviation of each feature in each transient fingerprint vector within the segment from its respective segment mean is calculated, and the absolute values ​​of the deviations corresponding to the same feature are cumulatively averaged to obtain the segment discrete value corresponding to each feature. Then, the segment discrete values ​​corresponding to each feature are normalized and weighted and summed to generate the fingerprint discrete value corresponding to the preset continuous action segment, where the weights are set to 0.09, 0.11, 0.11, 0.08, 0.08, 0.12, 0.10, 0.10, 0.10, and 0.11 respectively. For each preset continuous action segment, a rebound deterioration value is generated based on the number of contact rebounds and the total contact rebound time within the segment. In practice, for each preset continuous action segment, the number of contact rebounds and the total contact rebound time corresponding to each transient fingerprint vector within the segment are extracted. The segment average of the number of rebounds and the segment average of the total rebound time within the preset continuous action segment are calculated respectively. Then, the segment average of the number of rebounds is compared with the historical baseline value of the number of rebounds to obtain the deviation of the number of rebounds, and the segment average of the total rebound time is compared with the historical baseline value of the total rebound time to obtain the deviation of the total rebound time. Subsequently, the deviation of the number of rebounds and the deviation of the total rebound time are normalized respectively, and the average value is calculated to obtain the rebound deterioration value corresponding to the preset continuous action segment. The historical baseline values ​​for the number of rebounds and the total rebound time are set by extracting the number of contact rebounds and the total rebound time of the same switching device during the historical normal operation process, and averaging the multiple measurements of each. Generate an attraction decay value based on the attraction setup time; In practice, for each preset continuous action segment, the pull-in establishment time corresponding to each transient fingerprint vector within the segment is extracted, and the average pull-in establishment time within the preset continuous action segment is calculated. Then, the average pull-in establishment time is compared with the historical benchmark value of pull-in establishment time, and the difference between the average pull-in establishment time and the historical benchmark value is extracted as the pull-in attenuation value corresponding to the preset continuous action segment. When the average pull-in establishment time is not higher than the historical benchmark value of pull-in establishment time, the pull-in attenuation value is recorded as 0. The historical benchmark value for engagement setup time is set by extracting the engagement setup time corresponding to the same switching device during historical normal operation and averaging the multiple measurements. Based on the release decay time, a release hysteresis value is generated; In practice, for each preset continuous action segment, the release decay time corresponding to each transient fingerprint vector within the segment is extracted, and the average release decay time within the preset continuous action segment is calculated. Then, the average release decay time is compared with the historical baseline value of release decay time, and the difference between the average release decay time and the historical baseline value is extracted as the release hysteresis value corresponding to the preset continuous action segment. When the average release decay time is not higher than the historical baseline value of release decay time, the release hysteresis value is recorded as 0. The historical reference value for release decay time is set by extracting the release decay time corresponding to the same switching device during historical normal operation and averaging the multiple measurements. Based on the arc duration and arc intensity integral value within the section, an arc wear indication value is generated; In specific implementation, for each preset continuous action segment, the arc duration and arc intensity integral value corresponding to each transient fingerprint vector within the segment are extracted. The segment mean of arc duration and the segment mean of arc intensity integral value within the preset continuous action segment are calculated respectively. Then, the segment mean of arc duration is compared with the historical baseline value of arc duration to obtain the arc duration deviation, and the segment mean of arc intensity integral value is compared with the historical baseline value of arc intensity integral value to obtain the arc intensity integral value deviation. Subsequently, the arc duration deviation and the arc intensity integral value deviation are normalized respectively and weighted summation is performed to generate the arc wear indication value corresponding to the preset continuous action segment. The weights corresponding to the weighted summation are set to 0.4 and 0.6 respectively. The historical baseline values ​​for arc duration and arc intensity integral are set by extracting the arc duration and arc intensity integral values ​​corresponding to the same switching device during historical normal operation, and averaging the multiple measurements of each. Fingerprint difference value, fingerprint drift value, fingerprint dispersion value, rebound deterioration value, and arc wear indication value are all comprehensive evaluation indicators, while pull-in attenuation value and release hysteresis value are individual degradation characterization values. Among them, fingerprint difference value characterizes the overall degree of change in transient action response between adjacent actions of the same switching device within a preset continuous action section; fingerprint drift value characterizes the degree of continuous deviation of the transient action response of the same switching device along the action time sequence within a preset continuous action section; fingerprint dispersion value characterizes the degree of dispersion of transient action response fluctuation of the same switching device within a preset continuous action section; rebound deterioration value characterizes the degree of deterioration of contact rebound relative to the normal level; pull-in attenuation value characterizes the degree of slowing down of the pull-in establishment process relative to the normal level; release hysteresis value characterizes the degree of delay of the release attenuation process relative to the normal level; and arc wear indication value characterizes the degree of increase in wear risk due to arcing relative to the normal level. The fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value and arc wear indication value corresponding to each preset continuous action segment are arranged in the time sequence of the preset continuous action segment to generate an action memory attenuation sequence. The fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value, and arc wear indication value corresponding to each preset continuous action segment are arranged in a continuous sequence according to time. This completely preserves the degradation evolution trajectory of the same switching device during continuous operation, avoiding the loss of the sequential relationship between different preset continuous action segments. Based on this action memory attenuation sequence, subsequent life evolution determination and fault warning are established on the basis of continuous evolution information.

[0025] In this embodiment, the generation of the lifetime evolution indicator sequence specifically includes: Read the corresponding action memory decay sequence according to the switch device identifier, and extract the data corresponding to each preset continuous action segment; For each preset continuous action segment, a transient response degradation index is generated based on the fingerprint difference value, fingerprint drift value, and fingerprint discrete value. In practice, for each preset continuous action segment, the fingerprint difference value, fingerprint drift value and fingerprint discrete value are normalized and weighted summed to generate a transient response degradation index. The weights corresponding to the weighted summation are set to 0.35, 0.40 and 0.25 respectively. The transient response degradation index is a comprehensive evaluation index. This comprehensive evaluation index is not a single physical quantity, but a dimensionless characterization quantity obtained by combining the fingerprint difference value, fingerprint drift value and fingerprint discrete value after normalization. Among them, the fingerprint difference value represents the overall degree of change in transient action response between adjacent actions of the same switching device within a preset continuous action section; the fingerprint drift value represents the degree of continuous shift of the transient action response of the same switching device along the action time sequence within the preset continuous action section; and the fingerprint discrete value represents the degree of dispersion of transient action response fluctuation of the same switching device within the preset continuous action section. The weighted sum of the three values ​​comprehensively represents the overall level of transient action response degradation of the same switching device within the corresponding preset continuous action section. Based on the rebound deterioration value, the pull-in attenuation value, the release hysteresis value, and the arc wear indication value, a mechanism degradation index is generated; In practice, for each preset continuous action segment, the rebound deterioration value, pull-in attenuation value, release hysteresis value and arc wear indication value are normalized respectively, and then the normalization results are weighted and summed to generate the mechanism degradation index. The weights corresponding to the weighted summation are set to 0.25, 0.20, 0.20 and 0.35 respectively. The mechanism degradation index is a comprehensive evaluation index, a dimensionless characterization quantity obtained by combining the normalized values ​​of springback deterioration, pull-in attenuation, release hysteresis, and arc wear indication. Among them, the springback deterioration value characterizes the degree of deterioration of contact springback relative to the normal level, the pull-in attenuation value characterizes the degree to which the pull-in establishment process is slower than the normal level, the release hysteresis value characterizes the degree to which the release attenuation process is delayed relative to the normal level, and the arc wear indication value characterizes the degree to which the wear risk of arcing is increased relative to the normal level. The weighted sum of the above four values ​​comprehensively characterizes the overall deterioration level of the internal degradation mechanism of the same switching device within the corresponding preset continuous operation section. Based on the transient response degradation index and the mechanism degradation index, segment deviation value, segment coordination value and segment unidirectional enhancement value are generated; In specific implementation, for each preset continuous action segment, the absolute value of the difference between the transient response degradation index and the mechanism degradation index is calculated as the segment deviation value corresponding to that preset continuous action segment; the square root of the product of the transient response degradation index and the mechanism degradation index is extracted as the segment synergy value corresponding to that preset continuous action segment; the transient response degradation index corresponding to the current preset continuous action segment is compared with the transient response degradation index corresponding to the previous preset continuous action segment to obtain the transient response degradation index increment, and the mechanism degradation index corresponding to the current preset continuous action segment is compared with the mechanism degradation index corresponding to the previous preset continuous action segment to obtain the mechanism degradation index increment; when the transient response degradation index increment and the mechanism degradation index increment have the same sign, their product is extracted as the segment co-directional enhancement value corresponding to that preset continuous action segment; when the transient response degradation index increment and the mechanism degradation index increment have opposite signs, the segment co-directional enhancement value is recorded as 0, and the segment co-directional enhancement value corresponding to the first preset continuous action segment is recorded as 0; The segment synergy value is a comprehensive characterization measure used to characterize the degree to which transient response degradation and mechanistic degradation are enhanced together within the same preset continuous action segment. The segment deviation value characterizes the degree of difference between the transient response degradation index and the mechanistic degradation index within the same preset continuous action segment. The segment homogeneous enhancement value characterizes the degree to which the transient response degradation index and the mechanistic degradation index are enhanced synchronously in adjacent preset continuous action segments. The segment synergy value and the segment co-directional enhancement value are combined, and the segment deviation value is used for suppression correction to generate the segment lifetime evolution index. In specific implementation, for each preset continuous action segment, the segment coordination value is normalized to obtain the normalized segment coordination value; the segment unidirectional enhancement value is normalized to obtain the normalized segment unidirectional enhancement value; the normalized segment coordination value and the normalized segment unidirectional enhancement value are weighted and summed to obtain the segment coordination enhancement amount, where the weights are set to 0.6 and 0.4 respectively; the segment deviation value is normalized to obtain the normalized segment deviation value, and then 1 is subtracted from the normalized segment deviation value to obtain the deviation suppression coefficient; the segment coordination enhancement amount is multiplied by the deviation suppression coefficient to obtain the segment lifetime evolution index corresponding to the preset continuous action segment; The segment lifetime evolution index is a dimensionless comprehensive characterization of the degree of lifetime degradation evolution. Among them, the segment synergistic enhancement value characterizes the degree to which transient response degradation and mechanism degradation coexist and continuously enhance, and the deviation suppression coefficient characterizes the degree of suppression when the consistency between the two types of degradation is insufficient. The segment lifetime evolution index is obtained by multiplying the segment synergistic enhancement value and the deviation suppression coefficient. It reflects the overall degree of continuous evolution of the lifetime degradation of the switching device within the same preset continuous operation segment. The larger the segment lifetime evolution index, the higher the degree of continuous evolution of the lifetime degradation of the switching device, the greater the failure risk, and the shorter the remaining lifetime. The lifetime evolution index of each preset continuous operation segment is arranged in the time sequence of the preset continuous operation segment to generate the lifetime evolution indicator sequence of each switching device.

[0026] In this embodiment, the generation of the fault warning determination sequence specifically includes: Read the corresponding lifetime evolution indicator sequence and extract the segment lifetime evolution index corresponding to each segment according to the time sequence of the preset continuous action segments; Compare the segment lifetime evolution index corresponding to adjacent preset continuous action segments to extract segments with continuously rising index and segments with sudden increases in index. In practice, the lifetime evolution index of each segment corresponding to the same switching device is extracted according to the time sequence of the preset continuous operation segments. The lifetime evolution index of the current preset continuous operation segment is compared with the lifetime evolution index of the previous preset continuous operation segment. When the lifetime evolution index of the current segment is continuously higher than that of the previous segment and the number of consecutive occurrences reaches a preset threshold, the corresponding continuous preset operation segment is determined as the index continuously rising segment. When the difference between the lifetime evolution index of the current segment and the lifetime evolution index of the previous segment is higher than a preset surge threshold, the current preset continuous operation segment is determined as the index surge segment. In practice, the preset number threshold is taken as three consecutive preset continuous action segments, and the preset sudden increase threshold is taken as three times the average of the adjacent differences in the lifetime evolution index of the same switching device during its historical normal operation. By matching the continuous rise in the index with the sudden increase in the index, the potential fault warning targets and warning triggering sections can be identified. In practice, the corresponding continuously rising exponential segments and exponential spike segments are read according to the switch device identifier, and the start and end positions of each segment are extracted. The time position of each exponential spike segment corresponding to the same switch device is compared with each continuously rising exponential segment. When the start position of the exponential spike segment falls within the range of the corresponding continuously rising exponential segment, or is located in the adjacent preset continuous action segment after the end position of the corresponding continuously rising exponential segment, the switch device is identified as a latent fault warning object, and the exponential spike segment is identified as the warning trigger segment. When multiple exponential spike segments corresponding to the same switch device meet the above conditions at the same time, the exponential spike segment with the earliest start position is taken as the warning trigger segment. A continuously rising exponential range indicates that the switching device has shown a continuous degradation trend during a period of continuous operation, while a sudden increase in the exponential range indicates that this degradation trend has intensified at a certain point in time. When the starting position of the sudden increase in the exponential range falls within the range of the corresponding continuously rising exponential range, it means that the sudden increase occurs during the continuous degradation process and belongs to the same latent fault evolution chain. When the sudden increase in the exponential range is located within the adjacent preset continuous operation range after the ending position of the corresponding continuously rising exponential range, it means that the sudden increase immediately follows the continuous degradation and can still be regarded as a continuation of the aforementioned continuous degradation. Based on this, the switching device is identified as a latent fault warning target, and the sudden increase in the exponential range is identified as the warning trigger range, which can avoid misjudging isolated fluctuations as latent faults. Based on the cumulative value of the segment lifetime evolution index and the cumulative value of the segment lifetime evolution index increment within the warning triggering segment, the risk accumulation level of the corresponding switching device is determined; In specific implementation, for each warning triggering segment corresponding to a switching device, the segment life evolution index corresponding to each preset continuous action segment within the warning triggering segment is extracted, and the segment life evolution indexes are accumulated to obtain the cumulative value of the segment life evolution index; the positive increments of the segment life evolution index between adjacent preset continuous action segments within the warning triggering segment are extracted, and the positive increments are accumulated to obtain the cumulative value of the segment life evolution index increment; then the cumulative value of the segment life evolution index and the cumulative value of the segment life evolution index increment are normalized and weighted to generate a risk accumulation value, and the risk accumulation level of the corresponding switching device is determined based on the risk accumulation value, wherein the weights corresponding to the weighted sum are set to 0.7 and 0.3 respectively; In practice, the risk accumulation level is determined based on the range in which the risk accumulation value is located. Specifically, the preset range for low risk level is 0 to 0.35, the range for medium risk level is greater than 0.35 to 0.65, and the range for high risk level is greater than 0.65 to 1.00. Based on the warning triggering section, determine the fault type indication result; In practice, for each pre-warning triggering segment corresponding to a switching device, the fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value, and arc wear indication value corresponding to each preset continuous action segment within the pre-warning triggering segment are extracted. The values ​​of the same feature within the pre-warning triggering segment are cumulatively averaged to obtain the segment representative value corresponding to each feature. Then, the representative values ​​of each segment are compared, and the feature name corresponding to the segment representative value with the largest value is extracted. This feature name is used as the fault type indication result of the corresponding switching device. Fault types include contact rebound fault type, pull-in attenuation fault type, release hysteresis fault type, arc wear fault type, and transient response abnormal fault type. Fingerprint difference value, fingerprint drift value, and fingerprint discrete value together correspond to transient response abnormal fault type. The latent fault warning object, the warning triggering section, the risk accumulation level, and the fault type indication result are correlated to form a fault warning judgment sequence.

[0027] In this embodiment, the generation of fault warning results specifically includes: Read the warning triggering section, risk accumulation level and fault type indication results corresponding to each switching device in the fault warning judgment sequence; Extract the corresponding switching device identifier and generate the warning device identifier; Determine the early warning stage based on the level of risk accumulation; In practice, for each warning device identifier, the corresponding risk accumulation level is read; when the risk accumulation level is high, it is determined to be the emergency warning stage; when the risk accumulation level is medium, it is determined to be the warning stage; when the risk accumulation level is low, it is determined to be the attention stage. Use the cumulative risk level as the risk level; Based on the fault type indication results and the warning stage, generate maintenance prompt information; By associating the warning device identifier, warning stage, warning triggering section, risk level, and maintenance prompt information with the corresponding switch device identifier, the fault warning result can be obtained.

[0028] In this embodiment, the early warning stage includes the attention stage, the early warning stage, and the emergency early warning stage.

[0029] refer to Figure 2 The segment lifetime evolution trend curve uses the preset continuous operation segment number as the horizontal axis and the index value as the vertical axis to simultaneously characterize the transient response degradation index, the mechanism degradation index, and the segment lifetime evolution index. Both the transient response degradation index and the mechanism degradation index show a gradual upward trend throughout the overall process, indicating that the switching device undergoes an evolutionary process of simultaneous transient response performance degradation and the cumulative enhancement of internal degradation mechanisms during continuous operation.

[0030] Within the first to fifth preset continuous operation segments, the growth rates of various indices are relatively small, indicating a slow degradation phase. Within the sixth to ninth preset continuous operation segments, the segment lifetime evolution index shows a continuous increase, meeting the criteria for continuous rise, thus forming a segment with continuously rising indices. At the tenth preset continuous operation segment, the segment lifetime evolution index experiences a significant jump compared to the previous preset continuous operation segment, changing from a continuous growth state to a sudden increase state, thus forming a segment with a sudden increase in indices. This process indicates that the switching device gradually transitions from a slow degradation phase to an accelerated evolution phase, with intensified degradation characteristics appearing in local segments, thus providing a basis for identifying latent fault warning targets and determining warning trigger segments.

[0031] Example 1: To verify the feasibility of this invention in practice, it was applied to a low-voltage distribution box in a precision manufacturing workshop. This distribution box contains circuit breakers, AC contactors, and intermediate relays for controlling conveyor systems, workstation lighting, and localized power loads. The site is characterized by frequent start-stop cycles, dense load switching, and long continuous operating times. During long-term operation, maintenance personnel primarily rely on temperature rise, current, and voltage anomalies for inspection. However, in previous use, problems have arisen where, even when the temperature and current are still within acceptable ranges, the switching devices gradually develop issues such as slow engagement, increased contact rebound, and enhanced breaking arc, making it difficult to identify latent life-cycle faults in a timely manner. Therefore, an AC contactor of the same model was selected as the monitoring object in this distribution box. Transient response data during multiple opening, closing, engagement, and release processes were collected. The collected data includes action command time data, coil on / off current data, contact switching time data, current waveform data during the action process, rebound vibration waveform data, and arc response waveform data. The sampling interval was set to 1ms, and data was collected continuously for 36 days, resulting in 6120 valid action process segments. The transient response data was then processed sequentially for time synchronization, action segment extraction, noise removal, baseline correction, and device identification association to generate an action response sequence. Based on this sequence, the action delay duration, engagement setup duration, release attenuation duration, current surge peak value, current surge duration, waveform distortion, contact rebound count, total contact rebound duration, arc duration, and arc intensity integral value were calculated to form a set of action feature units. These feature units were then arranged and combined according to their action time sequence to construct a transient fingerprint vector, forming a historical transient fingerprint sequence for the switching device.

[0032] When applying this invention, five transient fingerprint vectors arranged sequentially according to the action time are used to form a preset continuous action segment. For each preset continuous action segment, fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value and arc wear indication value are calculated to generate an action memory attenuation sequence. In the initial monitoring phase, the switching device's action delay remained between 19ms and 21ms, the engagement setup time between 27ms and 29ms, the release decay time between 16ms and 18ms, the contact rebound count was 0 to 1, the total contact rebound time was 0ms to 3ms, the arc duration was 3ms to 4ms, the arc intensity integral value was stable between 0.18 and 0.24, the corresponding fingerprint difference value was 0.22 to 0.27, the fingerprint drift value was 0.18 to 0.21, the fingerprint dispersion value was 0.16 to 0.20, and the segment lifetime evolution index was 0.19 to 0.26. The lifetime evolution indicator sequence did not show any continuously increasing or abruptly increasing exponential segments. As the number of runs increased, in subsequent continuous monitoring, the action delay time increased to 28ms to 31ms, the engagement setup time increased to 36ms to 40ms, the release decay time increased to 23ms to 27ms, the number of contact rebounds increased to 3 to 4 times, the total contact rebound time increased to 10ms to 13ms, the arc duration increased to 8ms to 10ms, and the arc intensity integral value increased to 0.49 to 0.63; correspondingly, the fingerprint difference value increased to 0.58, the fingerprint drift value increased to 0.62, the fingerprint dispersion value increased to 0.47, the rebound deterioration value increased to 0.55, the engagement decay value increased to 0.39, the release hysteresis value increased to 0.36, and the arc wear indication value increased to 0.68. Further calculations showed that the transient response degradation index increased from 0.31 to 0.64, the mechanism degradation index increased from 0.29 to 0.66, and the segment lifetime evolution index was 0.48, 0.57, and 0.61 for three consecutive preset continuous action segments, and then suddenly increased to 0.79 in the subsequent adjacent preset continuous action segments, forming a continuous index increase segment and an index sudden increase segment. Based on this, the system identifies the switching device as a latent fault warning target and determines the corresponding exponential surge section as the warning trigger section. Further calculation yields a risk accumulation value of 0.74, classifying the risk accumulation level as high risk. Within the warning trigger section, after averaging the fingerprint difference value, fingerprint drift value, fingerprint dispersion value, rebound deterioration value, pull-in attenuation value, release hysteresis value, and arc wear indication value, the section corresponding to the arc wear indication value has the largest representative value of 0.61. Therefore, the fault type indication result is determined to be an arc wear fault type. The final output fault warning result includes the warning device identifier, emergency warning stage, warning trigger section, high risk level, and maintenance prompt information.Maintenance personnel shut down and inspected the target switching device based on the fault warning result, finding obvious ablation and edge melting on the contact surface. At this time, the circuit temperature was 45.8℃, still below the original 60℃ alarm threshold, and the operating current remained between 0.94 and 0.98 of the rated value. Traditional steady-state monitoring methods did not issue any abnormality alerts. Comparison with existing maintenance records shows that this invention provides early warning approximately 8 days before the occurrence of a manifest failure to operate, significantly earlier than the original method relying on steady-state parameters. Furthermore, the downtime for planned replacement of the device after the warning was 0.6 hours, while the average downtime after a manifest failure was previously 3.1 hours. This indicates that this invention can identify the evolution trend from normal to degradation and then to the pre-failure stage based on transient response changes during multiple operations, even before significant temperature rise and current abnormalities are observed. This effectively solves the problem of existing technologies' difficulty in timely detection of latent life-cycle faults in switching devices.

[0033] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A fault early warning method for identifying anomalies in distribution boxes, characterized in that, Includes the following steps: Acquire transient response data of switching devices in the distribution box during multiple opening, closing, energizing and releasing processes, and perform preprocessing to generate action response sequences; Based on the action response sequence, construct a set of action feature units; The action feature unit set is arranged and combined to construct a transient fingerprint vector, and then arranged in chronological order to generate a historical transient fingerprint sequence of the switching device; Cross-cycle comparison is performed on the historical transient fingerprint sequence of switching devices to form an action memory decay sequence; The action memory decay sequence is input into the life evolution determination process of the switching device to generate the segment life evolution index, and arranged in the time order of the preset continuous action segments to generate the life evolution indicator sequence. Based on the lifetime evolution indication sequence, the latent fault warning object, warning triggering section, risk accumulation level and fault type indication result are determined to form a fault warning judgment sequence. The fault warning result in the distribution box is determined based on the fault warning judgment sequence.

2. The fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The transient response data includes action command time data, coil on / off current data, contact switching time data, action process current waveform data, rebound vibration waveform data, and arc response waveform data.

3. The fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The preprocessing includes time synchronization, motion segment extraction, noise removal, baseline correction, and device identification association.

4. The fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the action feature unit set specifically includes: Read the action process segments corresponding to each single opening, closing, energizing, and releasing action response sequence; Based on the action command time data and the contact conduction switching time data, the action command issuance time and the contact conduction switching time of each action process segment are extracted, and the action delay time is calculated. Based on the coil current data, determine the coil energization start time, the coil current reaches the stable range, and the coil de-energization decay end time, and calculate the engagement setup time and release decay time. Extract the current peak point, impact duration, and waveform deviation from the current waveform data during the action process; extract the current impact peak value, current impact duration, and waveform distortion. By combining rebound vibration waveform data and contact conduction switching time data, contact rebound events are identified, the number of contact rebounds and the total contact rebound time are counted, and the arc start and end times and arc intensity change segments are extracted based on arc response waveform data, and the arc duration and arc intensity integral value are calculated. Summarize the action features to form a set of action feature units.

5. A fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the historical transient fingerprint sequence of the switching device includes: Extract the corresponding action feature unit set according to the switching device identifier, and sort each action feature unit according to the action time sequence; The sub-features in the action feature unit are combined to construct the transient fingerprint vector; Arrange the transient fingerprint vectors corresponding to the same switching device identifier in chronological order of action time to generate a historical transient fingerprint sequence of the switching device.

6. A fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the action memory decay sequence specifically includes: Read the historical transient fingerprint sequence of the corresponding switching device, and extract adjacent transient fingerprint vectors along the action time sequence to construct multiple preset continuous action segments; For each preset continuous action segment, extract the adjacent transient fingerprint vectors within the segment, compare the features of each dimension of the adjacent transient fingerprint vectors, and generate fingerprint difference values. In each preset continuous action segment, fingerprint drift value and fingerprint discrete value are generated based on the transient fingerprint vector within the segment. For each preset continuous action segment, a rebound deterioration value is generated based on the number of contact rebounds and the total contact rebound time within the segment. Generate an attraction decay value based on the attraction setup time; Based on the release decay time, a release hysteresis value is generated; Based on the arc duration and arc intensity integral value within the section, an arc wear indication value is generated; The fingerprint difference value, fingerprint drift value, fingerprint discrete value, rebound deterioration value, pull-in attenuation value, release hysteresis value and arc wear indication value corresponding to each preset continuous action segment are arranged in sequence to generate an action memory attenuation sequence.

7. A fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the lifetime evolution indicator sequence specifically includes: Read the corresponding action memory decay sequence according to the switch device identifier, and extract the data corresponding to each preset continuous action segment; For each preset continuous action segment, a transient response degradation index is generated based on the fingerprint difference value, fingerprint drift value, and fingerprint discrete value. Based on the rebound deterioration value, the pull-in attenuation value, the release hysteresis value, and the arc wear indication value, a mechanism degradation index is generated; Based on the transient response degradation index and the mechanism degradation index, segment deviation value, segment coordination value and segment unidirectional enhancement value are generated; The segment synergy value and the segment co-directional enhancement value are combined, and the segment deviation value is used for suppression correction to generate the segment lifetime evolution index. The lifetime evolution index of each preset continuous action segment is arranged in the time sequence of the preset continuous action segments to generate a lifetime evolution indicator sequence.

8. A fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the fault warning determination sequence specifically includes: Read the corresponding lifetime evolution indicator sequence and extract the segment lifetime evolution index corresponding to each segment according to the time sequence of the preset continuous action segments; Compare the segment lifetime evolution index corresponding to adjacent preset continuous action segments to extract segments with continuously rising index and segments with sudden increases in index. By matching the continuous rise in the index with the sudden increase in the index, the potential fault warning targets and warning triggering sections can be identified. Based on the cumulative value of the segment lifetime evolution index and the cumulative value of the segment lifetime evolution index increment within the warning triggering segment, the risk accumulation level of the corresponding switching device is determined; Based on the warning triggering section, determine the fault type indication result; The latent fault warning object, the warning triggering section, the risk accumulation level, and the fault type indication result are correlated to form a fault warning judgment sequence.

9. A fault early warning method for identifying anomalies in distribution boxes according to claim 1, characterized in that, The generation of the fault warning result specifically includes: Read the warning triggering section, risk accumulation level and fault type indication results corresponding to each switching device in the fault warning judgment sequence; Extract the corresponding switching device identifier and generate the warning device identifier; Determine the early warning stage based on the level of risk accumulation; Use the cumulative risk level as the risk level; Based on the fault type indication results and the warning stage, generate maintenance prompt information; By associating the warning device identifier, warning stage, warning triggering section, risk level, and maintenance prompt information with the corresponding switch device identifier, the fault warning result can be obtained.

10. A fault early warning method for identifying anomalies in distribution boxes according to claim 9, characterized in that, The early warning phase includes the attention phase, the early warning phase, and the emergency early warning phase.

Images