Power failure identification and detection system for marine electrical maintenance

Abstract

The application discloses a kind of power failure identification detection systems applied to ship electrical maintenance, including acquisition module, candidate identity generation module, identity disambiguation module, state determination module, attribution label output module and fault discrimination module.The application is by using the inherent physical characteristics that electromechanical switch action lags behind electromagnetic transient process, decouples and secondary correlation of electrical sampling data at the moment of fault triggering and the switch topology state converging after transient subsides on time axis, so that the determination of power supply identity is not affected by switch contact bounce and auxiliary contact jitter, and the finally generated candidate power supply identity set can correctly reflect the actual power supply path at the moment of fault, eliminating the omission or error of candidate identity caused by timing mismatch.The application solves the problem that the prior art does not have the ability to perceive the dynamic change of power supply identity and cannot eliminate the attribution deviation caused by topology identity drift.

Description

Technical Field

[0001] This invention belongs to the field of ship electrical system fault detection and identification technology, specifically a power fault identification and detection system applied to ship electrical maintenance. Background Technology

[0002] Shipboard electrical systems, operating as independently managed microgrids with limited capacity, typically employ a radial distribution network. This network utilizes segmented buses and interconnecting switches to facilitate flexible switching between multiple generators, shore power, and emergency power supplies. In fault identification and detection during electrical maintenance, existing methods often use the physical address of the data acquisition channel as the sole attribution label, directly associating the acquired voltage and current signals to the corresponding fixed feeder or equipment node. Based on this fixed topology, abnormal conditions are then attributed to the branch or upstream switch of that node. A different approach would directly associate the acquired electrical parameters to the corresponding fixed feeder or equipment node and perform fault diagnosis based on a pre-defined static topology. The most significant difference between shipboard power systems and land-based power grids lies in the frequent switching of power supply modes during their operation—such as the conversion between shore power and shipboard power, paralleling and disconnecting multiple generator sets, switching power supply via segmented busbars, and the activation of emergency power supplies. These conditions can all cause the actual power supply entities corresponding to the same physical acquisition channel to drift in different time windows. When the system still performs attribution analysis on the electrical data at the time of the fault according to the static topology relationship, there is no effective mechanism to distinguish between the changes in electrical characteristics caused by topology identity drift and the abnormal characteristics caused by actual equipment faults.

[0003] Due to frequent power supply mode switching during the operation of ship electrical systems, when a ship switches from shore power to ship power, or performs paralleling and disconnection operations of multiple generator sets, some critical loads are maintained by different power paths through sectionalizing switches or temporary jumper cables, while some non-critical loads are temporarily disconnected or transferred according to power management strategies. During such switching processes, the actual power supply entity corresponding to the same physical acquisition channel undergoes dynamic changes. If the fault identification system still performs cause analysis on electrical data according to static topology relationships, the following situation may occur: when an arcing ground fault occurs at a sectionalizing node, the system attributes the anomaly to the load. The branch circuit itself; voltage fluctuations caused by poor contact during temporary power transfer are misjudged by the system as feeder overcurrent; when an emergency generator is put into operation, the system judges the power supply as normal only if the contactor is stuck and it is not a non-emergency load that is expected to be energized, as long as the voltage amplitude and frequency meet the qualified standards. The above situations indicate that when the system regards the physical channel address as a permanent and unchanging attribution label, there is no effective mechanism to distinguish between the changes in electrical characteristics caused by topology identity drift and the abnormal characteristics caused by actual equipment failure. There is a misalignment between the fault attribution result and the actual physical power supply path, which affects the correct execution of subsequent fault selection and isolation logic.

[0004] Taking the arcing grounding of the tie node during the switch from shore power to ship power as an example, the system attributes this anomaly to the load branch based on the static topology, while the actual fault source is located at the cross-regional tie node, causing the fault selection logic to be misled to the wrong branch. Taking the contactor sticking during the operation of the emergency generator causing the non-emergency load to remain energized unexpectedly as an example, the system only detects that the voltage amplitude and frequency meet the qualified standards and judges the power supply to be normal, failing to identify the logical error that the load should not appear in the emergency power supply sequence. The common root cause of the above two types of failure scenarios is that the system mistakenly regards the address label of the physical channel as the identity label of the power supply entity. The decoupling that occurs between the two during the mode switching process is not perceived and compensated by any existing link, resulting in a systematic misalignment between the fault attribution result and the actual physical power supply path, thereby affecting the correct execution of the fault selection and isolation logic.

[0005] Among existing related patent technologies, such as the intelligent diagnostic model power quality disturbance tracing method based on a workflow platform disclosed in patent document (CN122085057A), the tracing logic of this patent is based on "known and accurate topology". However, in ship power systems, power supply mode switching (such as shore power / ship power conversion, generator paralleling / disconnection, bus interconnection and transfer) is often completed within milliseconds to seconds. The data refresh cycle of the scheduling system is usually on the order of seconds or even longer, with obvious communication delays and state update lags. When the physical topology has changed, but the scheduling system has not yet updated the switch state, the system will still build an undirected graph according to the "old static topology". At this time, the actual power supply entity corresponding to the physical acquisition channel has undergone "topology identity drift", but the system is unaware of this and directly applies the drifted electrical characteristics to the wrong topology node, which will inevitably lead to a completely wrong tracing direction. The core algorithm in this patent (energy flow direction inference, spatiotemporal clustering) is highly dependent on the two physical quantities of "adjacent nodes" and "electrical distance" based on a fixed topology. When topology identity drift occurs, the "electrical distance" and "adjacency" between the physical acquisition channel and other monitoring nodes have fundamentally changed. If the system fails to detect this change in real time and continues to use historically fixed electrical distances to calculate the attenuation gradient, it will lead to: abnormal jumps in gradient calculation results, which may be misjudged as new disturbance sources or interruptions in propagation paths; and the failure of spatiotemporal clustering conditions, resulting in the incorrect splitting of nodes that originally belonged to the same disturbance event, or the incorrect aggregation of nodes from different events. Although the "target hybrid model" in this patent can correct the propagation chain, its training dataset and inference logic are based on the premise that "topology relationships are correct." When topology identity drift occurs, due to changes in the internal impedance, short-circuit capacity, or harmonic background of the power supply, the voltage / current signals acquired by the same physical channel will exhibit drastic steps or distortions. Because this patent lacks a decoupling and feature differentiation mechanism for "topology switching events" and "equipment failure events," the target hybrid model is highly susceptible to mislearning or misjudging these "pseudo-abnormal features" caused by normal topology switching as "abnormal features generated by real equipment failures," thereby outputting incorrect root cause nodes.

[0006] Therefore, given the existing problems with the technology, it is necessary to enable the fault identification system to sense dynamic changes in power supply identity in order to eliminate attribution bias caused by topology identity drift. Summary of the Invention

[0007] The purpose of this invention is to provide a power fault identification and detection system for ship electrical maintenance, which solves the problem that the existing technology does not have the ability to sense dynamic changes in power supply identity and cannot eliminate attribution bias caused by topological identity drift;

[0008] The technical problem to be solved by this invention is: how to provide a power fault identification and detection system for ship electrical maintenance that has the ability to sense dynamic changes in power supply identity in order to eliminate attribution bias caused by topology identity drift.

[0009] The objective of this invention is achieved through the following technical solution: A power fault identification and detection system for ship electrical maintenance includes: Acquisition module: Acquires electrical sampling data packets, switch status sequences, and hardware channel identifiers of the monitoring nodes. The electrical sampling data packets include voltage waveform data and current waveform data before and after the fault reference time. Candidate identity generation module: Based on the steady-state convergence value of the switch state sequence within a preset time window after the fault reference time, backtrack to generate a set of candidate power supply identities corresponding to the monitoring node; Identity disambiguation module: For each candidate power supply identity, calculate the theoretical voltage phase based on its power supply side phase and path phase shift parameters, and generate a path consistency index based on the degree of difference between the measured voltage phase and the theoretical voltage phase; State determination module: Determines the current topology state as high-confidence valid state, multi-source fuzzy coupling state, or measurement anomaly state based on the path consistency index of each candidate power supply identity; Attribution tag output module: When the topology state is a high-confidence valid state, it outputs a flexible attribution tag containing the valid power supply identity and valid hold time; when the topology state is a multi-source fuzzy coupling state or a measurement anomaly state, it outputs a hardware channel identifier as an escape attribution tag. Fault identification module: Performs fault identification on the electrical sampling data packet based on the elastic attribution label or the escape attribution label, and outputs fault identification results with attribution information.

[0010] Furthermore, the switch state sequence is: the digital state of each circuit breaker and contactor auxiliary contact of the power supply path associated with the monitoring point, which is continuously collected at a preset scanning cycle; the switch state sequence is stored in the buffer in the form of transition event records, and each transition event record includes the switch number that transitioned, the time when the transition occurred, and the state value after the transition.

[0011] Furthermore, the candidate power supply identity set is generated as follows: starting from the fault reference time, a time window of a preset backtracking duration is extracted; all transition event records corresponding to each switch number within the time window are traversed, and the state value after the last transition of each switch number within the time window is taken as the steady-state convergence value of that switch number; if the time difference between the last transition time of the target switch and the end time when traversing to the end of the time window is less than the preset stability judgment duration, then the instantaneous state value of the target switch at the end time is taken as the steady-state convergence value, and a low confidence flag is added to the backtracking result corresponding to the target switch; based on the steady-state convergence values ​​of all switches and the electrical connection relationship between each switch, a conduction path from each power node to the monitoring node is searched, and the identity identifier of the power node corresponding to each searched conduction path is added to the candidate power supply identity set.

[0012] Furthermore, the measured voltage phase is obtained by extracting the phase angle of the fundamental positive sequence component by performing a one-cycle discrete Fourier transform on the voltage waveform data; the theoretical voltage phase is obtained by vector correction of the real-time phase of the power supply side of the candidate power supply identity according to the transformer phase shift angle on the power supply path corresponding to the candidate power supply identity; the path consistency index is generated in the following way: the phase angle difference between the measured voltage phase and the theoretical voltage phase is calculated, and the phase angle difference is normalized by adding or subtracting an integer period angle to make it fall into the standard angle range; the path consistency index is generated according to the deviation of the normalized phase angle difference from the preset reference tolerance angle, and the path consistency index decreases monotonically as the deviation increases.

[0013] Furthermore, the specific calibration process of the preset reference tolerance angle is as follows: calculate the ratio of the resistance component to the reactance component of the line impedance based on the cable type, cross-section and length between the monitoring node and the power supply side, determine the maximum steady-state phase angle difference under normal operating conditions based on the ratio, and use the sum of the maximum steady-state phase angle difference and the preset measurement error margin as the preset reference tolerance angle.

[0014] Furthermore, the specific method for determining the current topology state based on the path consistency index of each candidate power supply identity is as follows: the path consistency index of each candidate power supply identity is compared with a first preset threshold and a second preset threshold, wherein the first preset threshold is greater than the second preset threshold; when there is a unique candidate power supply identity whose path consistency index is greater than or equal to the first preset threshold, it is determined to be the high-confidence valid state, and the candidate power supply identity is determined to be the valid power supply identity; when the condition that there is a unique candidate power supply identity whose path consistency index is greater than or equal to the first preset threshold is not met, and there is at least one candidate power supply identity whose path consistency index is greater than or equal to the second preset threshold, it is determined to be the multi-source fuzzy coupling state; when the path consistency index of all candidate power supply identities is less than the second preset threshold, it is determined to be the measurement abnormal state.

[0015] Furthermore, the elastic attribution label is generated as follows: Three-phase instantaneous current values ​​within a preset number of power frequency cycles after the fault reference time are extracted from the current waveform data; the current value at the current sampling point is subtracted from the current value at the corresponding sampling point one power frequency cycle prior to generate a current increment sequence for each phase; a one-cycle discrete Fourier transform is performed on each phase current increment sequence to extract the fundamental amplitude and the second harmonic amplitude of each phase; an asymmetric distortion coefficient is generated based on the ratio of the fundamental amplitude to the second harmonic amplitude of each phase; a lower limit clamping process is applied to the denominator of the asymmetric distortion coefficient using a preset clamping constant; an attenuation factor is generated based on the degree of difference between the asymmetric distortion coefficient and the reference asymmetric distortion background value corresponding to the effective power supply identity, and the attenuation factor monotonically decreases as the degree of difference increases; the product of a preset maximum holding time and the attenuation factor is used as the effective holding time, and a lower limit clamping value is set for the effective holding time; the effective power supply identity and the effective holding time together constitute the elastic attribution label.

[0016] Furthermore, the reference asymmetric distortion background value is obtained by averaging the asymmetric distortion coefficients of the effective power supply identity under no-load and preset light-load conditions multiple times. Different effective power supply identities correspond to their respective reference asymmetric distortion background values ​​and are pre-stored in the system.

[0017] Furthermore, the specific method for performing fault identification on the electrical sampling data packet is as follows: When the flexible attribution label is output, the rated current value and rated short-circuit current multiple of the valid power supply identity are obtained by looking up a table according to the valid power supply identity in the flexible attribution label; the distance conversion factor from the current monitoring node to the power supply side of the valid power supply identity is obtained; the expected short-circuit current level is generated according to the rated current value, rated short-circuit current multiple and the distance conversion factor; the rated current carrying capacity of the feeder where the current monitoring node is located is obtained; a dynamic overcurrent start threshold is generated according to the comparison relationship between the expected short-circuit current level and the rated current carrying capacity; when the escape attribution label is output, a preset multiple of the rated current carrying capacity is used as a fixed overcurrent start threshold; the real-time effective current value in the electrical sampling data packet is compared and judged with the dynamic overcurrent start threshold or the fixed overcurrent start threshold, and the fault judgment result and the identity information in the attribution label currently used are output.

[0018] Furthermore, the generation of the candidate power supply identity set relies on a pre-stored ship power distribution network topology model. The topology model uses each power node, each monitoring node, and each switch node as vertices, electrical connection relationships as edges, and the current on / off state of each switch as the on / off attribute of each edge. The search method for the conduction path is as follows: starting from each power node, a traversal search is performed along the edges with the on / off attribute as the conduction state to obtain all paths that can reach the current monitoring node. When there are multiple starting power nodes corresponding to the searched paths, the identity identifiers of multiple power nodes are added to the candidate power supply identity set.

[0019] Compared with the prior art, the present invention has the following significant advantages: 1. By utilizing the inherent physical characteristic that the action of electromechanical switches lags behind the electromagnetic transient process, the electrical sampling data at the moment of fault triggering is decoupled and re-correlated with the switch topology state after the transient subsidence on the time axis. This makes the determination of power supply identity unaffected by switch contact bounce and auxiliary contact jitter. The final generated candidate power supply identity set can correctly reflect the actual power supply path at the moment of fault, eliminating the omission or error of candidate identity caused by timing mismatch. 2. By introducing the voltage phase angle consistency index as a continuous measure of path legality, the system can perform spatial domain verification of candidate power supply identities without pre-setting all possible topology combinations. When the path consistency falls into the multi-source fuzzy coupling range, the attribution logic actively abandons identity replacement and falls back to the hardware channel identifier, avoiding subsequent discrimination distortion caused by forced classification under unexpected topology states such as electromagnetic coupling or parasitic loops, so that the system still has predictable and definite behavioral boundaries when the topology is unknown. 3. By extracting the second harmonic distortion characteristics of the fault current as the modulation signal for the trust period of the identity tag, the effective retention time of the identity tag is automatically compressed as the nonlinearity of the fault intensifies. This negative feedback relationship ensures that the identity tag will not be locked in the wrong state for a long time under severe harmonic pollution or CT saturation conditions. A passive adaptive adjustment relationship is formed between the system's tracking response speed to topology changes and the degree of pollution of the measurement signal. 4. By using the final output attribution label as an index for fault discrimination, the dynamic overcurrent start threshold is automatically matched with the short-circuit capacity of the current power supply entity, eliminating the risk of false tripping or failure to trip caused by the protection settings not being updated synchronously after the power supply mode switch. At the same time, in fuzzy coupling and measurement anomaly escape modes, the basic protection capability is maintained with fixed settings. When the identity is reliable, dynamic matching criteria are enabled, so that the fault discrimination result is accompanied by the correct power supply identity information, providing an accurate attribution basis for subsequent fault selection and isolation operations. Attached Figure Description

[0020] Figure 1 This is a system structure block diagram of Embodiment 1 of the present invention; Figure 2 This is a flowchart of the system operation method according to Embodiment 1 of the present invention. Detailed Implementation

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

[0022] Example 1: As Figure 1 and Figure 2 As shown, the power fault identification and detection system of the present invention, applied to ship electrical maintenance, includes: Acquisition module: Acquires electrical sampling data packets, switch status sequences, and hardware channel identifiers of the monitoring nodes. The electrical sampling data packets contain voltage waveform data and current waveform data before and after the fault reference time. This embodiment sets up a data acquisition and preprocessing layer, which consists of synchronous phasor measurement units distributed and installed at each distribution board and feeder cabinet. Each measurement unit uses a synchronous clock source as a reference to synchronously sample the three-phase voltage and three-phase current of the monitoring node. The sampling rate is 40 sampling points per cycle, which is equivalent to a sampling frequency of 2000Hz under 50Hz power frequency conditions. Each sampling point is appended with a global timestamp. At the same time, the system continuously reads the digital status of each circuit breaker and contactor auxiliary contact of the power supply path of the associated monitoring point at a scanning rate of once per millisecond, i.e., 1000Hz.

[0023] The system allocates two types of circular buffers in memory; the electrical quantity buffer has a depth of 320 sampling points, corresponding to 8 power frequency cycles, and adopts first-in-first-out management; when writing a new sampling point, the system synchronously calculates the difference between the current sampling point and the corresponding sampling point 4 sampling cycles ago. This difference operation is used to suppress the steady-state fundamental component to highlight the transient change component; at the same time, the three-phase voltage sliding effective value is updated with the most recent 40 sampling points as a window; the switch quantity buffer only records switching events rather than continuous steady-state values: when a switch changes state in two adjacent scans, the system records an event entry containing the switch number, switching time, and post-switching state.

[0024] The system deploys a transient detection comparator on the write path of the electrical quantity buffer to monitor the absolute value of the voltage difference between each phase in real time. When the absolute value of the voltage difference between any phase exceeds the preset transient threshold for three consecutive sampling points, a transient event is determined to have occurred, and the current time is latched as the fault reference time. The transient threshold is set to 0.1 times the rated phase voltage peak value. This value is determined based on the statistical condition that the voltage fluctuation of the typical load switching voltage of the ship does not exceed 8% of the rated value, ensuring reliable triggering of real faults while avoiding interference from normal operation. After triggering, the system extracts a waveform segment from the electrical quantity buffer from the two power frequency cycles before to the three power frequency cycles after the fault reference time, i.e., from the first 40ms to the last 60ms, a total of 100 milliseconds, and encapsulates it together with the header information into an electrical sampling data packet. The first 40ms... The first 60ms is used to provide a steady-state background reference before the fault, and the last 60ms is used to cover the complete waveform of the transient fault process. This selection is based on the fact that the first two waves provide a steady-state reference before the fault, and the last three waves cover most of the first half of the transient wave and the subsequent complete cycle. The data packet header contains a hardware channel identifier, a fault reference time timestamp, and a trigger source type flag. The hardware channel identifier is a combination of the acquisition board slot number and the channel number, which remains unchanged throughout the system's lifecycle and serves as the physical anchor point for subsequent escape attribution. The system also extracts all transition event records within 200 milliseconds before and after the fault reference time from the switch quantity buffer to form a switch event record sequence. This extraction range is wider than the electrical quantity interception range to cover the electromechanical switch action delay and contact bounce duration.

[0025] Taking the shore power to ship power switching scenario as an example: During the contact bounce of the tie switch, a single-phase grounding occurs on the temporary tie bus. The system detects that the voltage difference of phase C reaches 0.12, 0.15 and 0.13 times the rated peak value at three consecutive sampling points, respectively, all exceeding the threshold. The fault reference time is latched and the waveform is captured. During this period, the auxiliary contact of the switch continues to bounce. The buffer records the alternating jump events. However, because the capture and extraction are parallel and independent, the switch bounce does not affect the correct capture of the waveform, and the data is completely preserved. At this point, the acquisition module completes the acquisition, buffering, triggering and encapsulation, and synchronously outputs the three data channels of electrical sampling data packet, switch quantity status sequence and hardware channel identifier to the candidate identity generation module.

[0026] Candidate identity generation module: Based on the steady-state convergence value of the switch state sequence within a preset time window after the fault reference time, backtrack to generate a set of candidate power supply identities corresponding to the monitoring node; After the acquisition module completes the encapsulation and output of electrical sampling data packets, switch status sequences, and hardware channel identifiers, the candidate identity generation module receives the above three data streams as input and performs a backtracking search operation for power supply identity.

[0027] In this topology model, the on / off attributes of each edge are refreshed at the beginning of each electrical sampling data packet processing cycle by reading the latest steady-state value after the transition from the switch quantity buffer. This refreshed value is used for routine topology state tracking when no transient events are triggered, for display on the system monitoring interface, and for status query under non-fault conditions. When a transient event is triggered and the candidate identity generation process begins, the system does not use this refreshed value as the sole basis for conduction path search, but instead uses the steady-state convergence value obtained by backtracking within the time window after the fault as the search benchmark. There is no mutual overriding relationship between the refreshed value and the backtracking convergence value; the former is stored in the edge attribute field of the topology model, while the latter is independently passed to the path search algorithm as the search parameter for this processing cycle. The model uses power nodes, monitoring nodes, and switch nodes as vertices, the electrical connection relationships between each vertex as edges, and the current on / off state of each switch as the on / off attribute of each edge. This topology model is entered once according to the electrical design drawings during the ship construction and commissioning phase. If temporary jumper cable changes occur subsequently, the operation and maintenance personnel manually refresh the connection relationships of the corresponding nodes in the configuration file before operation.

[0028] At the beginning of each electrical sampling data packet processing cycle, the on / off attributes of each edge in the above topology model are refreshed by reading the latest steady-state value after the transition from the switch quantity buffer. The refresh operation is independent of the steady-state convergence backtracking process of the candidate identity generation module. The former provides a search benchmark, and the latter performs path search based on this benchmark. When the candidate identity generation module uses the instantaneous state value as the search input due to the low confidence flag, the instantaneous value is only used for the on / off determination of this path search and is not written to the edge attribute field of the topology model. The on / off attributes stored in the topology model are always kept as the latest steady-state value confirmed by the stability determination. The conduction path result obtained by the search is used as the output of this processing cycle and is independent of the on / off attributes currently stored in the topology model. There is no mutual overwriting relationship between them.

[0029] Within the processing cycle corresponding to each frame of electrical sampling data packet, the module takes the fault reference time in the data packet header as the starting point and extracts a time window for backtracking the convergence state of the switch topology. The length of this time window is 150ms, and its value is determined based on the following engineering facts: the inherent operating time of commonly used frame circuit breakers in ships, including the mechanical transmission part, is usually no more than 80ms, and the contactor release time is no more than 60ms. 150ms covers the latest completion time of normal switch operation and leaves sufficient margin. Within this time window, the system traverses all transition event records in the switch state sequence, and takes the state value after the last transition of each switch number within the window as the steady-state convergence value of the switch. If a switch does not have any transition within the window, its instantaneous state value at the fault reference time is directly taken.

[0030] For target switches that still have special cases when traversing to the end of the window, that is, the time difference between the last transition time of the switch and the end time of the window is less than the preset stability determination time, this embodiment takes the instantaneous state value of the target switch at the end time of the window as the steady-state convergence value, and adds a low confidence flag to the backtracking result corresponding to the target switch; the time used to determine whether the switch state has fully converged is 20ms. The physical meaning of this time is: any transition that occurs within the last 20ms before the end of the above time window, the system cannot reliably determine whether the transition has truly stabilized, so it is marked by reducing the confidence rather than directly discarding the information.

[0031] Given that the switching quantity scan cycle is 1ms, the timestamps of all transition events have a quantization error of no more than ±0.5ms. When the time difference between the last transition time of a switch and the end time of the window falls within the range of 18ms to 22ms, the system marks the comparison result of the time difference with 20ms as a boundary ambiguity state and executes two branches simultaneously: using the instantaneous state value as the convergence value and adding a low confidence flag, while delaying the stability determination of the switch to the next processing cycle for re-evaluation. This boundary ambiguity processing is only activated when the time difference is near the preset stability determination time and does not participate in the steady-state convergence determination under normal operating conditions.

[0032] The system uses the steady-state convergence value of all switches as the current value of the on / off attribute of each edge, and performs a conduction path search on the topology graph model. This search does not change the pre-stored vertex and edge connection relationships in the topology graph model, but only replaces the on / off state of the edges with the steady-state convergence value to determine the path for this search. The search starts from each power node and performs a depth-first traversal along the edges with the on / off attribute in the on state to obtain all paths that can reach the current monitoring node. The identity of the starting power node corresponding to each searched path is added to the candidate power supply identity set. If the searched path corresponds to multiple power nodes, the identity of these power nodes is added to this set.

[0033] For example, within a 150ms window after the fault reference time, the tie switch's switching event record shows that it was closed in the last 100ms of the window, the generator outlet switch remained closed throughout the window, while the shore power inlet switch experienced a switching event in the last 30ms of the window and remained open. The system takes the closed states of the tie switch and generator outlet switch, and the open state of the shore power inlet switch as steady-state convergence values, and searches for two conduction paths in the topology graph—one from the main generator through the closed bus to the monitoring point, and the other from the auxiliary generator through the same bus to the monitoring point. Therefore, the identity identifiers of both generators are added to the candidate power supply identity set. If the last switching event of the shore power inlet switch occurs 15ms before the end of the window, and this time difference is less than the preset stability judgment duration of 20ms, the system adds a low confidence flag to the switch and takes its open state at the end of the window as the steady-state convergence value to continue the search. At this point, the candidate identity generation module outputs the candidate power supply identity set, the steady-state convergence values ​​of each switch, and the possible low confidence flags to the identity disambiguation module.

[0034] Identity disambiguation module: For each candidate power supply identity, calculate the theoretical voltage phase based on its power supply side phase and path phase shift parameters, and generate a path consistency index based on the degree of difference between the measured voltage phase and the theoretical voltage phase; The candidate power supply identity set and the switch steady-state convergence value output by the candidate identity generation module, together with the voltage waveform data in the electrical sampling data packet, are entered into the identity disambiguation module. The task of this module is to calculate the credibility of each power supply identity in the candidate set in order to distinguish between the real power supply path and the false path caused by the mis-transmission of switch quantity or temporary bridging.

[0035] This module extracts the instantaneous three-phase voltage value sequence of the first complete power frequency cycle after the fault reference time from the electrical sampling data packet. It performs a one-cycle rectangular window discrete Fourier transform on the sequence and extracts the phase angle of the fundamental positive sequence component as the measured voltage phase. The reason for selecting the first complete cycle after the fault rather than the data before the fault is that the system topology has entered a new transient process at the moment the fault occurs. The phase information before the fault cannot represent the electrical state of the actual power supply path after the fault. If the data before the fault is used, it will cause the phase reference to deviate, and thus make the subsequent path consistency index calculation lose its physical meaning.

[0036] The starting position of the aforementioned DFT window is based on the positive zero-crossing point where the three-phase voltage composite vector transitions from the negative half-cycle to the positive half-cycle. The negative zero-crossing point does not participate in the window starting position. The system first detects the moment when the three-phase voltage composite vector changes from a negative value to a positive value and crosses the zero point from the captured waveform data, and uses this moment as the starting point of the DFT window. If the positive zero-crossing point is not aligned with the sampling point, the sampling point closest to the positive zero-crossing point is selected as the starting point. The negative zero-crossing point is not processed for window starting position, logically eliminating the 180° phase shift caused by confusion of the zero-crossing direction.

[0037] When the above-mentioned interception operation is executed, the system simultaneously records the zero-crossing phase of the power frequency cycle at the fault reference time. If the interception endpoint from the fault reference time to 60ms after the fault fails to align with the zero-crossing point of the power frequency cycle or an integer multiple of the sampling point, the system performs zero-padding or truncation processing on the sampling points at the interception boundary to ensure that each segment of waveform data output to the downstream module takes the zero-crossing point of the power frequency cycle as the starting boundary. This boundary alignment operation does not affect the overall time span of the intercepted waveform data, but only adjusts the start and end sampling point indexes of the data segment during output.

[0038] For each power source in the candidate power supply identity set, the system receives the real-time voltage phase broadcast periodically from the power source's output measurement unit via the marine industry Ethernet. This broadcast period does not exceed 20ms, ensuring that the phase deviation from the fault occurrence time is controlled within an acceptable range. Simultaneously, the system queries the accumulated transformer phase shift angle values ​​stored in the configuration file along the path from the power source's output to the current monitoring point, subtracts these accumulated values ​​from the real-time phase on the power source side, and obtains the theoretical voltage phase at the monitoring point. For direct supply paths without transformers, this phase shift angle is zero.

[0039] After obtaining the measured voltage phase and the theoretical voltage phase, the system calculates the difference between the two to obtain the phase angle difference. Since the phase angle is periodic, the original calculation result of this difference may fall outside the standard angle range. The system performs a normalization operation on this difference by adding or subtracting an integer multiple of 360°, mapping it to the standard angle range of -180° to +180°. This normalization step ensures that the subsequent index calculation is not affected by the repetition of the angle period.

[0040] Based on this, the system generates a path consistency index according to the degree of deviation between the normalized phase angle difference and the preset benchmark tolerance angle. The index is calculated as follows: with 1 as the numerator and the sum of 1 and the square of the ratio of the normalized phase angle difference to the benchmark tolerance angle as the denominator. The index takes a maximum value of 1 when the phase angle difference is 0, takes 0.5 when the absolute value of the phase angle difference is equal to the benchmark tolerance angle, and approaches 0 when the absolute value of the phase angle difference is much greater than the benchmark tolerance angle. The physical meaning is clear and the boundary is smooth.

[0041] Regarding the source of the preset reference tolerance angle, this angle is calibrated based on the line impedance parameters between the monitoring node and the power supply side. Specifically, the system calculates the ratio of the resistive component to the reactive component of the line impedance based on the cable type, cross-section, and length. Based on this ratio, it determines the maximum steady-state phase angle difference that may occur at both ends of the line under normal operating conditions. Then, the sum of the maximum steady-state phase angle difference and the preset measurement error margin is used as the reference tolerance angle. For marine low-voltage power systems, this value usually falls within the range of 3° to 5°, depending on the cable length and cross-sectional specifications. In this embodiment, the actual calculated value after calibration is used as the standard, and the value is not fixed.

[0042] For example, the candidate power supply identity set of a certain monitoring node includes two power sources, namely the main generator and the auxiliary generator. For the main generator, the system calculates that its phase angle difference is 2°, and after substituting it into the above formula, the path consistency index is approximately 0.86. For the auxiliary generator, the system calculates that its phase angle difference is 12°, and after substituting it into the formula, the path consistency index is approximately 0.15. The consistency indices of the two candidate identities show a significant difference, and this difference will be used by the state determination module for subsequent topology state classification. At this point, the identity disambiguation module outputs each candidate power supply identity and its corresponding path consistency index to the state determination module.

[0043] State determination module: Determines the current topology state as high-confidence valid state, multi-source fuzzy coupling state, or measurement anomaly state based on the path consistency index of each candidate power supply identity; The candidate power supply identities and their corresponding path consistency indices output by the identity disambiguation module, together with the low confidence flags output by the candidate identity generation module, are entered into the state determination module. This module is responsible for converting the continuous values ​​of the path consistency indices into discrete topology state classification results.

[0044] This embodiment pre-configures two threshold parameters: a first preset threshold of 0.9 and a second preset threshold of 0.3; the first preset threshold is higher than the second preset threshold; the above values ​​are based on the functional characteristics of the path consistency index—when the reference tolerance angle is 5°, the phase angle difference corresponding to 0.9 is approximately 1.67°, which is within the reasonable fluctuation range of the phase difference between the two ends of the line under normal operating conditions; the phase angle difference corresponding to 0.3 is approximately 9.36°, which significantly exceeds the maximum steady-state phase angle difference that may occur in marine cables under normal load range; the range between the two thresholds corresponds to the phase angle difference range where unintended electromagnetic coupling may occur between asynchronous power sources.

[0045] The system first checks whether the candidate power supply identity set is empty. If it is empty, since there are no comparable path consistency indicators, it is directly determined to be a measurement anomaly and the subsequent three-stage judgment is not performed. The judgment result is then output to the attribution label output module. If the candidate power supply identity set is not empty, the system compares the path consistency indicators of each candidate power supply identity with the two thresholds mentioned above, and performs three-stage state classification based on the comparison results.

[0046] The first case: If and only if there is a path consistency index ≥ 0.9 for a unique power supply identity in the candidate set, the system will determine the topology state as a high-confidence valid state and identify the power supply identity as a valid power supply identity. This branch deals with the most common scenario where both switch quantity feedback and electrical quantity measurement are within the normal range. In this case, the system has a high degree of confidence in assigning the power supply identity.

[0047] The second scenario: When there is no single, unique path consistency index in the candidate set that reaches or exceeds 0.9—that is, including cases where no power supply identity reaches 0.9 but at least one reaches 0.3, and cases where two or more power supply identities simultaneously reach or exceed 0.9—the system determines the topology state as a multi-source fuzzy coupling state. The physical meaning of multiple candidates simultaneously reaching the first threshold is that the voltage phase of the current monitoring point is highly consistent with the theoretical phase of two or more power sources, indicating that the system may be in an unexpected dual-end power supply or reverse power supply state, making it impossible to reliably assign the monitoring point to a single power source. The physical scenario corresponding to this branch is usually that a low-resistance circulating current loop is formed between asynchronous power sources through unknown impedance, or the system is in a transition process of multiple generators operating asynchronously in parallel. In this state, the voltage phasor measurement value may be simultaneously affected by multiple power sources, making it impossible to reliably assign the monitoring point to a single power source.

[0048] The third scenario: When the path consistency index of all power supply identities in the candidate set is lower than 0.3, the system judges the topology state as a measurement anomaly. The physical causes corresponding to this branch may include the voltage transformer secondary circuit being disconnected, causing the measured phase angle to deviate completely from the actual value; incorrect phase sequence wiring, causing the phase sequence of the power supply side and the load side to not correspond; or the candidate power supply identity set itself failing to contain the real power supply path due to configuration errors.

[0049] Taking the scenario with two candidate identities, a main generator and an auxiliary generator, as an example: the path consistency index of the main generator is about 0.86, and that of the auxiliary generator is about 0.15; 0.86 is lower than 0.9 but higher than 0.3, and 0.15 is lower than 0.3. Based on this, the system determines the current topology state as a multi-source fuzzy coupling state. At this time, there is no unique path among the two candidate identities that reaches the high confidence threshold, indicating that although the deviation between the actual measured voltage phase and the theoretical phase of the main generator has not reached the level of complete mismatch, the deviation has exceeded the normal tolerance range, and there may be unexpected electromagnetic coupling or parasitic loops. This determination result will be transmitted to the attribution tag output module, triggering the escape attribution branch. At this point, the state determination module outputs its classification result, the effective power supply identity (if it exists), and all path consistency indices together.

[0050] Attribution tag output module: When the topology state is in a high-confidence valid state, it outputs a flexible attribution tag containing the valid power supply identity and valid hold time; when the topology state is in a multi-source fuzzy coupling state or a measurement anomaly state, it outputs a hardware channel identifier as an escape attribution tag. The topology status classification results, effective power supply identity, and consistency index of all paths output by the status determination module, together with the current waveform data in the electrical sampling data packet and the hardware channel identifier output by the candidate identity generation module, are entered into the attribution label output module; this module performs branch processing based on the topology status classification results.

[0051] If the state determination result is a multi-source fuzzy coupling state or a measurement abnormal state, the module directly reads the hardware channel identifier in the header of the electrical sampling data packet and outputs the identifier as an escape attribution label. The hardware channel identifier is a combination of the acquisition board slot number and the channel number, which does not change with the power supply topology and points to the same physical installation location throughout the system lifecycle. This branch does not need to perform any modulation calculations on the current waveform data and directly transmits the original physical address.

[0052] If the state determination result is a high-confidence effective state, the module enters the elastic attribution label generation process. This process extracts the three-phase instantaneous current values ​​of two power frequency cycles after the fault reference time from the current waveform data, and performs the following differential operation on each phase current sequence: subtract the current value of the current sampling point from the current value of the corresponding sampling point one power frequency cycle ago to obtain the current increment sequence of each phase. The physical purpose of this differential operation is to remove the steady-state load component so that the features extracted later can reflect the fault-addition component rather than the normal operating current.

[0053] A one-cycle discrete Fourier transform is performed on the current increment sequence of each phase to extract the fundamental amplitude and second harmonic amplitude of each phase. The system takes the maximum value of the second harmonic amplitude in the three phases and divides it by the sum of the maximum value of the fundamental amplitude in the three phases and a clamping constant. This ratio is defined as the asymmetric distortion coefficient. The value used for lower limit protection of the denominator is determined by the rated current of the feeder where the monitoring node is located, and is equal to 0.01 times the rated current. Its function is to prevent the fundamental amplitude from being too small under light load or no-load conditions, which would cause numerical divergence. The physical meaning of this coefficient is to quantify the degree of nonlinear distortion of the fault current waveform. The higher the proportion of the second harmonic, the more prominent the arc, ferromagnetic saturation or nonlinearity of the power electronic device in the fault, and the more serious the degree of harmonic pollution of the measurement signal.

[0054] The system internally stores a reference asymmetric distortion background value for each power supply and feeder combination. This background value is obtained by averaging the asymmetric distortion coefficients measured under no-load and light-load conditions (25% of rated load) for the power supply corresponding to that power supply identity. Different effective power supply identities correspond to their own reference values ​​and are pre-stored in the system storage area.

[0055] After obtaining the measured asymmetric distortion coefficient and the reference asymmetric distortion background value, the system calculates the difference between the two. If the difference is greater than 0, the attenuation factor is calculated, which is an exponential function with the natural constant as the base and the difference as the exponent. If the difference is less than or equal to 0, the attenuation factor is 1. The attenuation factor decreases monotonically as the measured distortion coefficient deviates from the background value. Its mathematical form determines that when the deviation reaches 0.5, the attenuation factor is approximately 0.37, and when the deviation reaches 1, the attenuation factor is approximately 0.14. This attenuation relationship is used to control the effective retention time of the identity tag: the larger the deviation, the higher the nonlinearity of the current fault, the greater the possibility of the phase and amplitude information of the measurement system being contaminated, and the identity tag should be refreshed more quickly.

[0056] The effective holding time is calculated by multiplying the preset maximum holding time by the attenuation factor. The maximum holding time of the identity tag is 3 seconds, which covers the longest duration from triggering to completion of a typical ship mode switch. The system further sets a lower limit protection value for the effective holding time, which is 200ms, to prevent the identity tag from frequently changing at a frequency exceeding the system's response capability under conditions where the attenuation factor is too small due to an extremely severe fault. The clamping operation ensures that even under conditions where the attenuation factor is extremely small due to an extremely severe fault, the identity tag will not change at a high frequency exceeding the system's response capability. Finally, the effective power supply identity and the modulated and clamped effective holding time together constitute the elastic attribution tag output.

[0057] The effective retention time limit starts at the moment the transient event is triggered. After the time limit is reset to zero, if the system does not detect a new transient event, the power supply identity tag of the monitoring node remains in an untrusted state. The subsequent fault discrimination module operates with a fixed overcurrent start threshold until the next transient event is triggered, at which point the candidate identity generation and subsequent processing flow of all modules are re-executed. If a new transient event is triggered before the time limit is reset to zero, the effective retention time limit is recalculated starting from the new trigger moment. The old tag is simultaneously invalidated from the new event trigger moment. The system only retains the newly calculated elastic attribution tag as the input basis for the subsequent fault discrimination module. The invalidation of the old tag does not involve the reverse modification of the output discrimination results, but only affects the switching of the discrimination benchmark after the new trigger moment.

[0058] Taking an auxiliary generator as an example, the baseline asymmetric distortion background value for this power supply is 0.04. In a certain fault, the measured asymmetric distortion coefficient reaches 0.13, with a difference of 0.09 and an attenuation factor of approximately 0.84. The effective holding time is 3 seconds multiplied by 0.84, which is 2.52 seconds. If in another fault, the measured asymmetric distortion coefficient reaches 0.54, with a difference of 0.50 and an attenuation factor of approximately 0.37, the effective holding time is compressed to 1.11 seconds. Based on the state determination result, this module selects the corresponding branch and outputs the elastic attribution label or escape attribution label to the fault discrimination module.

[0059] Fault identification module: Performs fault identification on electrical sampling data packets based on flexible attribution labels or escape attribution labels, and outputs fault identification results with attribution information.

[0060] The elastic attribution tag or escape attribution tag output by the attribution tag output module, together with the real-time effective current value in the electrical sampling data packet, enters the fault discrimination module; this module performs differentiated discrimination benchmark determination logic according to the type of attribution tag received.

[0061] The system maintains a power parameter mapping table, which is stored in the same configuration file as the topology model on which the candidate identity generation module depends. This table is loaded into memory during system initialization. Each record in the table contains a unique power source identifier, rated current value, and rated short-circuit current multiple. The rated short-circuit current multiple is defined as the ratio of the expected effective value of the short-circuit current at the power source outlet under a three-phase metallic short circuit to its rated current. This value is calculated from the generator's transient reactance and system impedance. The configuration file also stores a distance conversion factor for each monitoring node. This factor is the ratio of the total cable impedance between the current monitoring node and its power source to the total impedance of the power source outlet to the farthest busbar within its power supply range. This factor is calculated based on the cable type and length during the electrical design phase and then entered all at once.

[0062] If the received tag is a flexible attribution tag, the system parses the valid power supply identity field in the tag, uses the identity as an index to query the power supply parameter mapping table, and obtains the corresponding rated current value and rated short-circuit current multiple. The rated current value, rated short-circuit current multiple, and distance conversion factor of the current monitoring node are multiplied to obtain the expected short-circuit current level. The rated current carrying capacity of the feeder where the current monitoring node is located is obtained, and the expected short-circuit current level is compared with the rated current carrying capacity. If the expected short-circuit current level is greater than the rated current carrying capacity, the dynamic overcurrent start threshold is determined as follows: the rated current carrying capacity is used as the base value, plus 0.6 times the difference between the expected short-circuit current level and the rated current carrying capacity. If the expected short-circuit current level is less than the rated current carrying capacity, it indicates that the monitoring point is extremely far from the power supply and the line impedance has decayed to the point that the short-circuit current level is lower than the normal operating current. In this case, 1.1 times the rated current carrying capacity is taken as the minimum start threshold. When the expected short-circuit current level is equal to the rated current carrying capacity, the same minimum threshold is used as the starting threshold, which is 1.1 times the rated current carrying capacity, to ensure that the starting threshold is always higher than the upper limit of the fluctuation of the normal load current. If the received tag is an escape attribution tag, it means that the current topology is in the fuzzy coupling or measurement abnormality range. The system will no longer perform the above dynamic calculation and will directly use 1.2 times the rated current carrying capacity of the feeder where the current monitoring node is located as the fixed overcurrent starting threshold.

[0063] When the distance conversion factor is less than 0.05, the monitoring node is determined to be in the near zone of the power supply outlet. Under this condition, once a short circuit occurs, the measured current will rise sharply to several times the value of the dynamic threshold, and the output delay of the fault discrimination module will be forcibly shortened to the minimum action delay of the protection corresponding to the power supply. This forced shortening only applies to monitoring nodes with a distance conversion factor of less than 0.05 and does not change the discrimination logic of other nodes.

[0064] After the start-up threshold is determined, the system writes the threshold value into the comparison register of the fault discrimination comparator, replacing the original fixed value. The comparator compares the real-time current effective value with the written threshold value point by point. When the real-time current effective value continuously exceeds the threshold value and reaches the preset delay, it outputs a fault trigger signal. The delay value is a fixed value to avoid the inrush current at the moment of start-up. The fault discrimination result output after triggering includes a fault type label. The label is determined to be a single-phase grounding, two-phase short circuit, or three-phase short circuit type based on the sequence component characteristics in the current waveform. At the same time, the identity information in the currently used attribution label is appended to the header of the discrimination result.

[0065] Taking the actual operating data of a certain monitoring node as an example, this node belongs to the main generator power supply domain. The rated current of the main generator is 2000A, and the rated short-circuit current multiple is 6. The distance conversion factor of the current monitoring node is 0.3, and the calculated expected short-circuit current level is 3600A. The rated current carrying capacity of the feeder is 400 amperes. The expected short-circuit current level is significantly greater than the rated current carrying capacity. The dynamic overcurrent start threshold is 400 plus the difference between 3600 and 400, multiplied by 0.6, which is 2320. This threshold value is much higher than the rated load current but lower than the maximum short-circuit current, which can distinguish between serious faults in the near area and minor faults in the far area. If the output tag switches to escape mode, the threshold degenerates to a fixed value of 480A, and the system operates in the basic overcurrent protection mode. At this point, the fault discrimination module outputs the fault discrimination result with power supply identity attribution information to the upper-level monitoring and protection logic unit of the system bus.

[0066] When the above modules execute in concert, after a transient event is triggered, the system first uses the electromechanical hysteresis characteristics of switching actions to backtrack and converge the topology, thereby generating a set of candidate power supply identities. Then, it filters each candidate identity based on the spatial consistency of the voltage phase angle, excluding candidates with path consistency below a threshold, thus narrowing the scope of attribution analysis to physically reliable power supply paths. For states where path consistency falls into an ambiguous range, the system abandons identity replacement and retreats to the absolute physical identifier of the hardware channel to avoid forced classification when the topology is unclear, which could lead to subsequent discrimination distortion. For identities that pass the consistency check, the system reverse-modulates their trust level based on the degree of asymmetric distortion of the fault current. The stronger the nonlinearity of the fault, the shorter the retention period of the identity tag, forcing the system to reassess the attribution earlier in the event of severe harmonic pollution. Through the series operation of the above four links of backtracking generation, phase angle screening, fuzzy escape and distortion negative feedback, the attribution process of electrical sampling data at the time of fault no longer depends on the pre-fixed physical channel address, but is driven by physical quantities of three orthogonal dimensions: switching quantity, voltage phasor and current distortion. Ultimately, the dynamic overcurrent start threshold is automatically matched with the physical capacity of the current power supply entity. This closed-loop mechanism can maintain the deterministic boundary of the attribution logic under mode switching and abnormal topology conditions, so that the fault judgment result is accompanied by the correct power supply identity information.

[0067] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents. Various modifications, additions, or similar substitutions made by those skilled in the art to the described specific embodiments, as long as they do not depart from the structure of the invention or exceed the scope defined by the claims, should fall within the protection scope of the present invention.

Claims

1. A power fault identification and detection system for ship electrical maintenance, characterized in that, include: Acquisition module: Acquires electrical sampling data packets, switch status sequences, and hardware channel identifiers of the monitoring nodes. The electrical sampling data packets include voltage waveform data and current waveform data before and after the fault reference time. Candidate identity generation module: Based on the steady-state convergence value of the switch state sequence within a preset time window after the fault reference time, backtrack to generate a set of candidate power supply identities corresponding to the monitoring node; Identity disambiguation module: For each candidate power supply identity, calculate the theoretical voltage phase based on its power supply side phase and path phase shift parameters, and generate a path consistency index based on the degree of difference between the measured voltage phase and the theoretical voltage phase; State determination module: Determines the current topology state as a high-confidence valid state, a multi-source fuzzy coupling state, or a measurement anomaly state based on the path consistency index of each candidate power supply identity; Attribution tag output module: When the topology state is a high-confidence valid state, it outputs a flexible attribution tag containing the valid power supply identity and valid hold time; when the topology state is a multi-source fuzzy coupling state or a measurement anomaly state, it outputs a hardware channel identifier as an escape attribution tag. Fault identification module: Performs fault identification on the electrical sampling data packet based on the elastic attribution label or the escape attribution label, and outputs the fault identification result with attribution information.

2. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The switch state sequence is: the digital state of each circuit breaker and contactor auxiliary contact of the power supply path associated with the monitoring point, which is continuously collected at a preset scanning cycle; the switch state sequence is stored in the buffer in the form of jump event records, and each jump event record includes the switch number that jumps, the time when the jump occurs, and the state value after the jump.

3. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The candidate power supply identity set is generated as follows: starting from the fault reference time, a time window of a preset backtracking duration is extracted; all transition event records corresponding to each switch number within the time window are traversed, and the state value after the last transition of each switch number within the time window is taken as the steady-state convergence value of that switch number; if the time difference between the last transition time of the target switch and the end time when traversing to the end of the time window is less than the preset stability judgment duration, then the instantaneous state value of the target switch at the end time is taken as the steady-state convergence value, and a low confidence flag is added to the backtracking result corresponding to the target switch; based on the steady-state convergence values ​​of all switches and the electrical connection relationship between each switch, a conduction path from each power node to the monitoring node is searched, and the identity identifier of the power node corresponding to each conduction path is added to the candidate power supply identity set.

4. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The measured voltage phase is obtained by extracting the phase angle of the fundamental positive sequence component by performing a one-cycle discrete Fourier transform on the voltage waveform data; the theoretical voltage phase is obtained by vector correction of the real-time phase of the power supply side of the candidate power supply identity according to the transformer phase shift angle on the power supply path corresponding to the candidate power supply identity; the path consistency index is generated in the following way: the phase angle difference between the measured voltage phase and the theoretical voltage phase is calculated, and the phase angle difference is normalized by adding or subtracting integer period angles to make it fall into the standard angle range; The path consistency index is generated based on the degree of deviation between the normalized phase angle difference and the preset reference tolerance angle. The path consistency index decreases monotonically as the degree of deviation increases.

5. The power fault identification and detection system for ship electrical maintenance according to claim 4, characterized in that, The specific calibration process of the preset reference tolerance angle is as follows: calculate the ratio of the resistance component to the reactance component of the line impedance based on the cable type, cross-section and length between the monitoring node and the power supply side, determine the maximum steady-state phase angle difference under normal operating conditions based on the ratio, and use the sum of the maximum steady-state phase angle difference and the preset measurement error margin as the preset reference tolerance angle.

6. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The specific method for determining the current topology status based on the path consistency index of each candidate power supply identity is as follows: the path consistency index of each candidate power supply identity is compared with a first preset threshold and a second preset threshold, wherein the first preset threshold is greater than the second preset threshold. When the path consistency index of a unique candidate power supply identity is greater than or equal to the first preset threshold, it is determined to be the high-confidence valid state, and the candidate power supply identity is determined to be the valid power supply identity. When the path consistency index of the unique candidate power supply identity is not satisfied and is greater than or equal to the first preset threshold, and there is at least one candidate power supply identity whose path consistency index is greater than or equal to the second preset threshold, it is determined to be the multi-source fuzzy coupling state. When the path consistency index of all candidate power supply identities is less than the second preset threshold, it is determined to be the measurement abnormal state.

7. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The elastic attribution label is generated as follows: Three-phase instantaneous current values ​​within a preset number of power frequency cycles after the fault reference time are extracted from the current waveform data; the current value at the current sampling point is subtracted from the current value at the corresponding sampling point one power frequency cycle prior to generate a current increment sequence for each phase; a one-cycle discrete Fourier transform is performed on each phase current increment sequence to extract the fundamental amplitude and the second harmonic amplitude of each phase; asymmetric distortion coefficients are generated based on the ratio of the fundamental amplitude to the second harmonic amplitude of each phase; and a lower limit clamping process is applied to the denominator of the asymmetric distortion coefficients using a preset clamping constant. An attenuation factor is generated based on the degree of difference between the asymmetric distortion coefficient and the reference asymmetric distortion background value corresponding to the effective power supply identity, and the attenuation factor decreases monotonically as the degree of difference increases. The effective holding time is the product of the preset maximum holding time and the attenuation factor, and a lower limit clamping value is set for the effective holding time; the effective power supply identity and the effective holding time together constitute the flexible attribution label.

8. The power fault identification and detection system for ship electrical maintenance according to claim 7, characterized in that, The reference asymmetric distortion background value is obtained by taking the average value of the asymmetric distortion coefficient after multiple measurements of the effective power supply identity under no-load and preset light-load conditions. Different effective power supply identities correspond to their respective reference asymmetric distortion background values ​​and are pre-stored in the system.

9. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The specific method for fault identification of the electrical sampling data packet is as follows: When the elastic attribution label is output, the rated current value and rated short-circuit current multiple of the valid power supply identity are obtained by looking up a table according to the valid power supply identity in the elastic attribution label; the distance conversion factor from the current monitoring node to the power supply side of the valid power supply identity is obtained; the expected short-circuit current level is generated according to the rated current value, the rated short-circuit current multiple and the distance conversion factor; the rated current carrying capacity of the feeder where the current monitoring node is located is obtained; a dynamic overcurrent start threshold is generated according to the comparison relationship between the expected short-circuit current level and the rated current carrying capacity; when the escape attribution label is output, a preset multiple of the rated current carrying capacity is used as a fixed overcurrent start threshold. The real-time effective current value in the electrical sampling data packet is compared and judged using the dynamic overcurrent start threshold or the fixed overcurrent start threshold, and the fault judgment result and the identity information in the currently used attribution tag are output.

10. The power fault identification and detection system for ship electrical maintenance according to claim 1, characterized in that, The generation of the candidate power supply identity set relies on a pre-stored ship power distribution network topology model. The topology model uses each power node, each monitoring node, and each switch node as vertices, electrical connections as edges, and the current on / off state of each switch as the on / off attribute of each edge. The search method for the conduction path is as follows: starting from each power node, a traversal search is performed along the edges whose on / off attribute is on to obtain all paths that can reach the current monitoring node. When there are multiple starting power nodes corresponding to the searched paths, the identity identifiers of multiple power nodes are added to the candidate power supply identity set.

Images