A shore power switching device state evaluation method based on multi-electric parameter measurement

Abstract

The present application belongs to the field of port shore power system operation and maintenance technology, and discloses a shore power switching device state evaluation method based on multi-electric parameter measurement, comprising: triggering evaluation through device state signals, generating absolute space-time anchor points, and driving synchronous sampling; adjusting the periodic sampling frequency, capturing multi-electric parameter data, and generating multi-electric parameter process slice data set; calculating data dynamic health degree, forming integrated data sequence, and generating high-reliability multi-electric parameter fusion data set with complete pedigree label; performing system-level physical law audit, locating fault source, generating abnormal equipment component list, extracting equipment component multi-parameter to generate collaborative degradation trajectory, and forming system-level physical audit report; extracting fault characteristics, matching historical fault case library to recommend maintenance scheme, generating operation and maintenance decision work order containing operation and maintenance linkage instruction for multi-berth parallel field, and performing closed-loop self-evolution through double-path update mechanism to form evaluation management closed-loop link.

Description

Technical Field

[0001] This invention relates to the field of port shore power system operation and maintenance technology, and more specifically, to a method for assessing the condition of shore power transfer devices based on multi-parameter measurement. Background Technology

[0002] With the intelligent upgrading of port shore power facilities and the advancement of green port construction, multi-berth parallel shore power systems have become the mainstream power supply mode for large ports. However, the large number of devices, strong parameter coupling, and complex environmental interference (such as salt spray and vibration) in multi-berth parallel operation scenarios place higher demands on the condition assessment of shore power transfer devices. Traditional condition assessment methods struggle to balance accuracy and timeliness, resulting in delayed deterioration warnings and low efficiency in fault handling.

[0003] Existing shore power transfer device status assessment schemes still have some shortcomings: First, they lack the ability to adapt to multi-berth parallel operation scenarios, resulting in a serious disconnect between assessment and operation and maintenance decisions. Most existing schemes are designed for independent operation scenarios of a single berth, failing to consider the load coupling and equipment linkage characteristics of multi-berth parallel operation. They can only output isolated fault alarm information, lacking operation and maintenance linkage commands such as load transfer and collaborative monitoring, leading to a disconnect between assessment and operation and maintenance decisions and reliance on manual intervention. Second, data collection and processing lack precise control mechanisms and a full-link data health assessment process, making them prone to misjudgments and omissions due to data interference, thus failing to provide a reliable basis for operation and maintenance decisions. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for assessing the condition of a shore power transfer device based on multi-parameter measurement, comprising:

[0005] S1: Trigger evaluation through equipment status signals, generate and broadcast absolute spatiotemporal anchor points, drive all monitoring nodes to sample synchronously; at the same time, combine historical equipment health and environmental data to dynamically adjust the periodic sampling frequency, capture multi-electrical parameter data, and integrate to generate multi-electrical parameter process slice datasets.

[0006] S2: Based on the multi-parameter process slice dataset, the dynamic health of the data is calculated according to the progressive evaluation rules to form an integrated data sequence; through three-level filtering and weighting, a high-confidence multi-parameter fusion dataset with complete lineage labels is generated;

[0007] S3: Perform system-level physical law audit on high-confidence datasets, locate fault sources, generate a list of abnormal equipment components; extract multi-parameters of equipment components to generate collaborative degradation trajectories, and integrate them to form a system-level physical audit report;

[0008] S4: Extract fault characteristics based on physical audit reports, match and recommend maintenance solutions from the historical fault case library, and generate maintenance decision work orders containing maintenance linkage instructions for multi-berth parallel operation scenarios;

[0009] S5: Based on the execution results of operation and maintenance decision work orders, package them into standard case archives, execute closed-loop self-evolution through a dual-path update mechanism, and form an evaluation and management closed-loop link.

[0010] Furthermore, the generation method of the absolute spatiotemporal anchor point includes:

[0011] Continuously capture mechanical status signals and electrical operation signals of key component monitoring nodes of the shore power unit at the berth, use them as equipment status signals, and decode and map them into equipment operating status;

[0012] The device's operating status, after real-time decoding and mapping, is compared and determined with a pre-set trigger judgment rule base.

[0013] Once the trigger evaluation conditions are met, an absolute spatiotemporal anchor point containing a unique identifier and a precise timestamp is generated based on the standard time and broadcast to all monitoring nodes via high-priority communication.

[0014] Furthermore, the generation method of the multi-parameter process slice dataset includes:

[0015] Each monitoring node synchronously starts sampling based on an absolute spatiotemporal anchor point, and simultaneously combines historical equipment health data with environmental data to construct a two-dimensional judgment matrix, dynamically adjusting the periodic sampling frequency of core components; then, using the adjusted sampling frequency, it captures full-cycle multi-electrical parameter data of the triggering event.

[0016] The collected multi-electrical parameter data is aligned with anchor reference time and filtered for validity. The filtered valid data is then integrated into a structured dataset based on time series and parameter type to form a multi-electrical parameter process slice dataset.

[0017] Furthermore, the integrated data sequence is formed in the following ways:

[0018] Based on the multi-parameter process slice dataset, the health of each data point is comprehensively calculated according to the progressive evaluation rules of sensor self-test, environmental dynamic threshold adjustment, redundancy reading consistency verification and physical limit paradox check, combined with weight allocation. The health level label and associated information are attached to the data points to form an integrated data sequence.

[0019] Furthermore, the generation method of the high-confidence multi-electrical parameter fusion dataset includes:

[0020] Based on the integrated data sequence, the raw multi-electrical parameter data of each monitoring node are filtered in three levels according to the health level label to retain the effective core data;

[0021] The effective core data are weighted and fused with health status as the weight, and a complete genealogical label is added to the fused data to form a highly reliable multi-electrical parameter fusion dataset.

[0022] Furthermore, the method for generating the list of abnormal equipment components includes:

[0023] Based on a high-reliability multi-electrical parameter fusion dataset, the system matches corresponding audit rules through a system-level physical law audit rule base, and performs matching power distribution node current balance audit, system power conservation audit, impedance matching and power flow audit by audit dimension to identify data anomalies or equipment anomalies.

[0024] For any abnormal results discovered during the audit, we will locate and trace the source, analyze the causes by combining the data from the entire process of traceability, mark the level of abnormality, and integrate and generate a list of abnormal equipment and components.

[0025] Furthermore, the system-level physical audit report is generated in the following ways:

[0026] Based on the list of abnormal equipment components, multi-dimensional parameters of abnormal equipment components are extracted from a high-confidence multi-electrical parameter fusion dataset, and the starting time point of single parameter degradation is marked.

[0027] Furthermore, through multi-parameter degradation time difference analysis and correlation modeling, the characteristics of collaborative degradation are determined, and the collaborative degradation trajectory of equipment components is generated;

[0028] By integrating audit results and full-process traceability data with collaborative degradation trajectory as the core, a system-level physical audit report is constructed.

[0029] Furthermore, the method of recommending repair solutions by matching historical fault case databases includes:

[0030] Based on the system-level physical audit report, multi-dimensional fault features are extracted according to the three-level architecture of equipment, fault and environment tracing, and standardized fault feature vectors are generated through quantitative coding processing.

[0031] Based on the fault feature vector, invalid cases are filtered out in the pre-set historical fault case library using an exact matching strategy, and the similarity between the current and historical case feature vectors is calculated using a fuzzy matching strategy to select high-matching and medium-matching cases.

[0032] Based on the repair solution with the highest similarity among the highly matched cases, and taking into account the repair steps, current operating conditions and maintenance resources of the medium-matched cases, an adaptation adjustment plan is generated and integrated into a repair solution recommendation report.

[0033] Furthermore, the generation method of the operation and maintenance decision work order includes:

[0034] Based on the maintenance plan recommendation report and the system-level physical audit report, for multi-berth parallel operation scenarios, the current operation data of the multi-berth parallel operation equipment cluster is collected and correlated to assess the operation status and load distribution characteristics of the equipment cluster and predict the feasibility of operation and maintenance linkage.

[0035] For multi-berth parallel operation scenarios with linkage feasibility, operation and maintenance linkage instructions are constructed through pre-set load transfer logic, process monitoring thresholds and safety interlocks, and maintenance plans are associated to form operation and maintenance decision work orders.

[0036] Furthermore, the packaging into a standard case archive, and the closed-loop self-evolutionary mechanism implemented through a dual-path update mechanism, includes:

[0037] Obtain data on the execution process of operation and maintenance decision work orders, combine it with fault verification results and post-repair retest data, perform structured classification and association, package it into standard cases and store it in the historical fault case library;

[0038] Based on the cases entered into the database, the evaluation system is updated through a dual-path update mechanism that combines auditing and matching rule optimization with health parameter adjustment, and validity verification is performed. The system takes effect after the verification is successful.

[0039] The technical effects and advantages of the present invention, which provides a method for assessing the condition of shore power transfer devices based on multi-parameter measurement, are as follows:

[0040] This invention focuses on multi-berth parallel shore power scenarios, and constructs a closed-loop management system for the entire process from data acquisition to operation and maintenance iteration, achieving the goals of accurate equipment status assessment, rapid fault handling, and continuous optimization of the assessment system.

[0041] First, by synchronous sampling at absolute spatiotemporal anchor points and dynamic frequency adjustment, and by dynamically adjusting the sampling frequency in conjunction with equipment health and environmental data, a unified time reference is established for all monitoring nodes to ensure the consistency of time sequence of multi-parameter data.

[0042] Secondly, by integrating progressive data health assessment with three-level filtering, from sensor self-inspection, environmental threshold adaptation, redundancy consistency verification to physical limit paradox checks, the credibility of data is quantified across the entire chain, generating a high-credibility dataset with spectral labels, eliminating the impact of interfering data on the assessment results, and improving the accuracy of the assessment.

[0043] Then, through system-level physical law auditing and collaborative degradation trajectory construction, auditing is carried out based on physical rules such as Kirchhoff's laws and the law of conservation of energy to locate the fault source and generate component collaborative degradation trajectories, so as to realize traceable analysis of the fault cause, avoid the interpretability problem caused by the algorithm black box, and provide a clear basis for fault handling.

[0044] Next, through fault feature matching and multi-berth operation and maintenance linkage instruction design, maintenance solutions are recommended based on the historical case library. Load transfer logic, process monitoring thresholds and safety interlock instructions are designed for the parallel operation characteristics of multiple berths. Root causes, verification steps and solutions are integrated to generate structured operation and maintenance decision work orders, realizing the automation and linkage of fault handling, significantly shortening operation and maintenance response time and reducing the risk of multi-berth collaborative operation.

[0045] Finally, by optimizing the full-process data archiving and dual-path evaluation system, the operation and maintenance execution data, verification results, and retest data are packaged into standard cases and stored in the database. Through the dual-path iterative evaluation system of rule weight optimization and health parameter adjustment, the closed-loop management of "collection-analysis-decision-execution-optimization" is realized, which significantly improves the long-term adaptability of the system.

[0046] This invention fundamentally solves the pain point of shore power transfer device status assessment in multi-berth parallel operation scenarios by constructing a rule-driven full-process assessment system. It is applicable to various port multi-berth shore power systems, significantly improving the accuracy of status assessment, the timeliness of operation and maintenance decisions, and the robustness of system operation. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of a method for assessing the condition of a shore power transfer device based on multi-electrical parameter measurement according to the present invention.

[0048] Figure 2 This is a schematic diagram of the core health assessment process of a single data point progressive health status in a shore power transfer device status assessment method based on multi-electrical parameter measurement according to the present invention.

[0049] Figure 3 This is a schematic diagram of the status assessment system for shore power transfer devices based on multi-electrical parameter measurement according to the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1:

[0051] Please see Figure 1 and Figure 2 As shown in this embodiment, a method for assessing the condition of a shore power transfer device based on multi-parameter measurement includes:

[0052] S1: Trigger evaluation through equipment status signals, generate and broadcast absolute spatiotemporal anchor points, drive all monitoring nodes to sample synchronously; at the same time, combine historical equipment health and environmental data to dynamically adjust the periodic sampling frequency, capture multi-electrical parameter data, and integrate to generate multi-electrical parameter process slice datasets.

[0053] S2: Based on the multi-parameter process slice dataset, the dynamic health of the data is calculated according to the progressive evaluation rules to form an integrated data sequence; through three-level filtering and weighting, a high-confidence multi-parameter fusion dataset with complete lineage labels is generated;

[0054] S3: Perform system-level physical law audit on high-confidence datasets, locate fault sources, generate a list of abnormal equipment components; extract multi-parameters of equipment components to generate collaborative degradation trajectories, and integrate them to form a system-level physical audit report;

[0055] S4: Extract fault characteristics based on physical audit reports, match and recommend maintenance solutions from the historical fault case library, and generate maintenance decision work orders containing maintenance linkage instructions for multi-berth parallel operation scenarios;

[0056] S5: Based on the execution results of operation and maintenance decision work orders, package them into standard case archives, execute closed-loop self-evolution through a dual-path update mechanism, and form an evaluation and management closed-loop link.

[0057] The methods for generating absolute spacetime anchor points include:

[0058] The monitoring scope is defined to cover the key components of all shore power transfer devices at all berths, including core switch components, power supply branch sensor nodes, etc., to ensure that there are no blind spots or monitoring blind areas in the entire area;

[0059] The device continuously and in parallel captures equipment status signals from various monitoring points of key components of the device in real time, including both mechanical status signals and electrical operation signals.

[0060] The mechanical status signals include circuit breaker opening and closing status signals (passive dry contact signals), contactor closing and opening status signals (NPN type open collector output), and load switching actuator action signals (analog 0-10V feedback); the electrical operation signals include real-time sampled values ​​of three-phase voltage, three-phase current, and zero-sequence current, with signal types of 4-20mA analog or RS485 digital signals (Modbus-RTU protocol).

[0061] The captured equipment status signals undergo signal preprocessing: high-frequency electromagnetic interference (such as spike pulses generated by the start and stop of port cranes) is suppressed through the built-in RC low-pass filter circuit (cutoff frequency 10Hz), and then the signal is smoothed through a first-order low-pass digital filter algorithm (filter coefficient = 0.8) to eliminate instantaneous jitter. For power frequency interference, a 50Hz notch filter is enabled to attenuate power frequency and harmonic interference (2nd and 3rd harmonics). A threshold for the signal amplitude range is set (such as voltage signal 0-500V, current signal 0-600A). Signals exceeding the range are marked as invalid and discarded to avoid abnormal values ​​interfering with subsequent judgments.

[0062] The preprocessed equipment status signals are decoded in real time to extract core information and map it to the equipment operating status: the dry contact signals of circuit breakers, contactors, etc. are decoded into clear states such as closed or open, attracted or disconnected; the analog signals of current and voltage are linearly converted into actual physical quantities and signal acquisition timestamps are added; the operating status of each monitoring point is mapped and updated in real time through a preset signal value and equipment status mapping table.

[0063] The device's operating status, after real-time decoding and mapping, is compared and judged one by one with the trigger evaluation conditions in the preset trigger judgment rule base.

[0064] The triggering rule base includes two categories of conditions: mechanical triggering conditions and electrical triggering conditions.

[0065] Mechanical triggering conditions are defined as non-redundant switching of the mechanical state of equipment (such as a circuit breaker changing from closed to open, or a contactor changing from closed to open).

[0066] The electrical triggering condition is defined as the change in electrical parameters exceeding a preset sensitivity threshold (e.g., a sudden change of 25V in voltage within 100ms, exceeding the ±22V threshold; a sudden change of 120A in current within 50ms, exceeding the ±100A threshold).

[0067] If any trigger evaluation condition is met, it is marked as a suspected trigger event and enters a two-layer secondary verification: First, confirm that the duration of the trigger signal is ≥50ms (configurable). If it is an instantaneous pulse signal (duration <50ms), it is judged as an interference signal and the trigger request is rejected. Then, perform consistency verification on the associated signals (e.g., the circuit breaker opening and closing signal and the corresponding branch current signal need to change synchronously. If the circuit breaker opens but the current does not return to zero, it is judged as a signal abnormality and the evaluation is not initiated).

[0068] After the secondary verification is passed, the trigger event is officially confirmed to be effective. Low-priority tasks of routine signal monitoring (such as periodic printing of the status matrix) are immediately suspended, and dedicated additional computing power is allocated to start the dedicated evaluation process.

[0069] After the event is confirmed, the built-in Beidou or GPS time synchronization tool is invoked to obtain nanosecond-level standard time by searching for satellites.

[0070] Based on the acquired standard time, an absolute spatiotemporal anchor point is generated, which includes anchor point ID (ID refers to a unique identifier), timestamp, trigger event type, trigger source device ID, and main control unit ID. At the same time, the uniqueness of the anchor point is ensured by using an auto-incrementing serial number method to avoid duplicate anchor point identifiers for different trigger events.

[0071] The absolute time anchor point is broadcast synchronously to all monitoring nodes in the shore power transfer device in the form of an integrated data packet via an industrial-grade Ethernet communication link according to a high-priority transmission protocol. After receiving the data packet, each monitoring node immediately parses and extracts the time reference information, completes the preliminary calibration of the local clock with the anchor point time, and sends a confirmation signal of successful anchor point reception to the main control unit. After receiving confirmation signals from all nodes, the main control unit determines that the construction of the absolute time anchor point is complete, thus laying a unified time reference for subsequent synchronous sampling.

[0072] Methods for generating multi-parameter process slice datasets include:

[0073] After the absolute spatiotemporal anchor point is constructed, a synchronous sampling preparation command is sent to all normally synchronized nodes. After receiving the command, each monitoring node starts sampling synchronously based on the absolute spatiotemporal anchor point. If a monitoring node fails to start sampling normally, it immediately reports an abnormality and marks the node as a sampling failure node. Subsequent data will not be included in the fusion analysis.

[0074] Synchronously access the device's historical health data (the average of the assessment results over the past 30 days) and environmental data (salt spray, concentration, vibration acceleration) to construct a two-dimensional judgment matrix of health and environment, in order to determine the sampling frequency adjustment strategy;

[0075] In the dual-dimensional judgment matrix, the historical health status of the equipment, environmental level (salt spray, humidity, vibration) and periodic sampling frequency (core components) are the rows, and the specific parameters are the columns;

[0076] Example two-dimensional decision matrix for default parameters (can be adjusted based on actual circumstances and expert experience):

[0077] [Equipment historical health level label (specific value); Environmental level (salt spray, humidity, vibration); Periodic sampling frequency (core components)];

[0078] [Excellent (≥0.9); Good (salt spray ≤60%RH, vibration ≤2g); 24 hours / time];

[0079] [Good (0.7-0.9); Fair (salt spray 60%-85%RH, vibration 2-3g); 12 hours / time];

[0080] [General (0.5-0.7); Harsh (salt spray > 85%RH, vibration > 3g); 8 hours / time];

[0081] [Poor (<0.5); Any environmental level; 4 hours / time];

[0082] It should be noted that for the first evaluation (or when a new device is put into use or when there is no historical health record after the device is initialized), the default health value of 0.6 (corresponding to the general health level) is uniformly used as the basis for calculating the two-dimensional judgment matrix. This means that the default value will not lead to gradual degradation and missed detection due to being too high, nor will the default value be too low to cause oversampling and increase resource consumption.

[0083] When it is determined that the sampling frequency needs to be adjusted based on the historical health data of the equipment and environmental data, a frequency adjustment instruction is sent to all core component monitoring nodes, including the new sampling period and the effective time of the adjustment (effective from the next sampling period). After receiving the instruction, the monitoring nodes record the adjustment parameters and provide feedback confirmation to ensure that the sampling frequency of all core component nodes is adjusted synchronously.

[0084] With the adjusted sampling frequency, the full-cycle multi-electrical parameter data of the trigger event is captured completely, and the range of multi-electrical parameters collected covers three-phase voltage, three-phase current, zero-sequence current, harmonics (2nd-20th), insulation resistance, impedance of key components and temperature rise data;

[0085] For the multi-electrical parameter data of all monitoring nodes collected, high-frequency electromagnetic interference is further eliminated by wavelet threshold denoising algorithm (wavelet basis is db4, threshold is calculated according to adaptive strategy); then the sampling data of each node is uniformly converted into JSON format to ensure that the data format of different nodes and different channels is consistent, and then data with sampling timestamp and anchor time deviation > 100ms and data with amplitude exceeding physical limit are eliminated.

[0086] For the retained sampled data, perform anchor point reference time alignment and validity filtering;

[0087] Among them, anchor point reference time alignment: based on the absolute spatiotemporal anchor point, time offset compensation is performed, the local timestamp of each data is extracted, and the offset from the anchor point time is calculated (offset = local timestamp - anchor point time).

[0088] The data timestamps are calibrated based on the offset. After calibration, the data timestamps of all nodes and all channels are uniformly aligned to the anchor time base.

[0089] Validity screening: Extract the dataset from anchor time -10ms to anchor time +50ms to ensure coverage of the complete dynamic process before and after the triggering event; calculate the deviation value for the synchronous data of redundant acquisition channels (such as two current sensors deployed on the same component), and mark data with a deviation >2% as inconsistent; only retain valid data with a deviation ≤2%; remove dead data with no change in amplitude (such as fixed values ​​caused by sensor failure) to ensure that the data can reflect the true operating status of the equipment.

[0090] The filtered valid data are structured and integrated according to time series and parameter type to form a multi-parameter process slice dataset, which includes basic metadata (unique anchor point identifier, trigger event type, trigger source device ID, data acquisition time range, and list of nodes participating in the acquisition), time series data matrix (a two-dimensional time series data matrix with timestamps as rows and parameter types as columns, with each cell representing the parameter acquisition value at the corresponding time point), and auxiliary information (time alignment log, data removal log, and environmental data snapshots (salt spray, humidity, and vibration data during the acquisition period)).

[0091] The methods for forming integrated data sequences include:

[0092] Based on the multi-parameter process slice dataset, the time series data matrix is ​​split according to the acquisition node and parameter type to form an independent data sequence of single node and single parameter. At the same time, the association mapping between data point, sensor ID and environmental conditions is established to provide a basic association for subsequent health calculation.

[0093] The dynamic health of each data point is comprehensively calculated based on the progressive evaluation rules of sensor self-test, environmental dynamic threshold adjustment, redundancy reading consistency verification and physical limit paradox check, combined with weight allocation.

[0094] Specifically, the first step: sensor self-test

[0095] Perform a full-dimensional self-check on the sensor corresponding to each data point to ensure that the sensor itself is in normal condition: check the feedback result of the sensor's built-in self-check command. If the self-check result is a failure (any self-check item is abnormal), directly mark the health of all data points of the sensor as 0, add a sensor fault mark, and trigger a sensor fault alarm simultaneously; if the self-check result is normal, proceed to the next step of calculation.

[0096] Step 2: Adjusting dynamic environmental thresholds (to adapt to complex port environments)

[0097] Based on data-linked environmental data (salt spray concentration, humidity, vibration acceleration), and according to preset environmental level classification rules (preset by expert experience), the health assessment thresholds for each parameter are dynamically adjusted (the original thresholds are the allowable deviation range of the equipment's rated parameters).

[0098] Taking the current fluctuation threshold as an example, for a good environmental level, the corresponding threshold adjustment rule is the original threshold (±2% of the rated current), and no adjustment is made; for a normal environmental level, the corresponding threshold is the original threshold × 1.3 (±2.6% of the rated current); for a severe environmental level, the corresponding threshold is the original threshold × 1.5 (±3% of the rated current).

[0099] If it is the first assessment and there is no historical environmental data, the threshold will be adjusted according to the general environmental level by default; the adjusted threshold will be recorded along with the data points to ensure that the health calculation is traceable.

[0100] Step 3: Redundant sensor reading consistency verification

[0101] For redundant sensors (≥2 sensors of the same model) deployed in the parameter configuration, perform a reading consistency check:

[0102] Verification logic: Calculate the reading deviation of redundant sensors at the same timestamp (reading deviation = |sensor A reading - sensor B reading| ÷ rated parameter value × 100%).

[0103] Health deduction rules:

[0104] Reading deviation ≤2%: judged as consistent, and the health status of redundant sensor data retains the base value (1.0).

[0105] 2% < Deviation value ≤ 5%: judged as slight inconsistency, the health status of all involved redundant sensor data is reduced to 0.7;

[0106] Deviation value > 5%: judged as serious inconsistency, the health of all involved redundant sensor data is reduced to 0.3, a redundancy contradiction mark is added, and a data review alarm is triggered;

[0107] If there are no redundant sensors for the parameters, skip this step and retain the health result from the previous step.

[0108] Step 4: Physical limit paradox check (avoiding invalid data)

[0109] Based on the physical operating principles of electrical equipment, a limit paradox check is performed on each data point to eliminate invalid data that violates physical laws.

[0110] Includes power limits (excluding abnormal data such as negative power (non-generating conditions) and power exceeding 150% of the equipment's rated power), voltage and current limits (excluding out-of-range data such as voltage <0V or >500V (power conditions of 220V and 380V) and current <0A or >600A (500A rated conditions)), and impedance limits (excluding abnormal data such as impedance of critical components <0Ω or >200% of rated impedance).

[0111] Judgment rule: If the data violates any physical limit, the health of the data point will be directly reduced to 0.2, and a physical limit anomaly mark will be added. It will only be used for trend monitoring and will not participate in the core analysis.

[0112] The final health score is calculated based on the weighted proportions of each progressive step, with the specific weight allocation as follows (which can be slightly adjusted according to port operation and maintenance needs):

[0113] Sensor self-test (basic weight): 40% (0.4 points for normal self-test, 0 points for failure);

[0114] Environmental threshold adaptation (adaptation weight): 20% (data within the adjusted threshold receives 0.2 points, data outside the threshold receives 0 points).

[0115] Redundancy consistency (redundancy weight): 25% (consistency scores 0.25 points, slight inconsistency scores 0.175 points, and severe inconsistency scores 0.075 points).

[0116] Physical Limit (Limit Weight): 15% (0.15 points for no paradox, 0.03 points for paradox).

[0117] Calculation method: Health score = base weight score + adaptation weight score + redundancy weight score + extreme weight score (total score range 0-1).

[0118] Based on the final calculated health score, a health score label is attached to each data point along with the associated sensor ID, environmental level, and verification anomaly marker (if any). The health score data is then aligned with the original multi-electrical parameter data by timestamp to form an integrated data sequence of data value, health score, and associated information.

[0119] Methods for generating high-reliability multi-electrical parameter fusion datasets include:

[0120] Based on the integrated data sequence, the raw multi-electrical parameter data of each monitoring node are filtered in three levels according to the health level label to retain the effective core data;

[0121] Specifically, based on the health level, for the original multi-electrical parameter data, data with a health level ≤ 0.2 in the poor category are directly removed and do not participate in any analysis. If the removal rate is > 30%, re-collection is triggered. Data with a health level > 0.2 in the poor category are marked as trend monitoring data, used only for long-term degradation trend analysis, stored separately, and not included in the core fusion dataset. Data in the general and good categories are retained as valid core data for subsequent weighted fusion and analysis, and proceed to the next fusion process.

[0122] The filtering is performed by iterating through the time series point by point. After each monitoring node is filtered, a node-level filtering log is generated, which records the node ID, total number of data points, data percentage of each level, timestamp of the removed data points, and the amount of trend monitoring data.

[0123] The effective core data is weighted and integrated based on health status. Specifically:

[0124] The weight value is completely consistent with the health value. For redundant sensor data with the same timestamp and the same parameter, an additional sensor accuracy weight is introduced. The final weight = health weight + sensor accuracy weight (total score ≤ 1).

[0125] A weighted average fusion algorithm is used to calculate the fusion value for all valid core data with the same timestamp and the same parameter according to the final weight. The formula is: Fusion value = Σ (data value × corresponding weight) ÷ Σ (corresponding weight).

[0126] If there is only one valid core data for the same parameter, the data value is directly used as the fusion value, and the weight is recorded as 1.0; if there is no valid core data for the same timestamp, linear interpolation of the data of the five timestamps before and after is used to supplement it, with an interpolation error ≤2%; if it exceeds 2%, the interpolation is marked as abnormal.

[0127] For the merged data stream, 20% of the timestamps are randomly selected and compared with the original valid core data to calculate the fusion deviation (fusion deviation = |fusion value - original data value| ÷ original data value × 100%).

[0128] If the fusion deviation is ≤3%, the fusion is considered valid; if 3% < deviation ≤ 5%, the fusion deviation is marked as slightly abnormal, the fusion result is retained and recorded; if the deviation > 5%, the fusion is considered to have failed, the fusion calculation for this parameter is re-executed, and if it still fails, a manual review alarm is triggered.

[0129] A complete phylogenetic label is attached to the fused data. The complete phylogenetic label includes sensor combination ID (the concatenation of all sensor IDs participating in the fusion), fusion weight information (health weight of each sensor and final weight), health level (the comprehensive health level of the fused data (taking the lowest level of the data participating in the fusion)), environmental condition snapshot (salt spray concentration, humidity, and vibration level at the time of sampling), and data processing mark (normal, interpolation supplementation, slight abnormality of fusion deviation).

[0130] The fused data is structured and integrated according to time series and parameter type to construct a high-confidence multi-electrical parameter fusion dataset, which includes basic metadata, time series fusion data matrix (with timestamps as rows and parameter types as columns, each cell containing a fusion value and a complete genealogy label), and auxiliary information.

[0131] The methods for generating a list of abnormal equipment parts include:

[0132] Based on a high-reliability multi-electrical parameter fusion dataset, the entire process traceability data is associated, including metadata (collection nodes, environmental snapshots) of the multi-electrical parameter process slice dataset, integrated data sequences, and logs synchronously recorded during the filtering and fusion process (including information such as rejection rate and fusion deviation).

[0133] Based on the electrical operating principle of shore power transfer devices, a system-level audit rule base with pre-set electrical parameters, physical laws, and verification thresholds is established. The core rule is defined as follows (structure: audit dimension - applicable physical law - core verification logic - verification threshold (which can be fine-tuned through expert experience) - associated parameters):

[0134] Distribution node current balance - Kirchhoff's Current Law (KCL) - The sum of all inflow currents at the same node = the sum of outflow currents - Current imbalance ≤ 5% - Three-phase current, zero-sequence current;

[0135] System power conservation - law of energy conservation - total input power = total output power + system power loss - power deviation ≤ 3% - three-phase voltage, current, and power factor;

[0136] Impedance matching verification - Ohm's law (U=IR) - Voltage of key components ÷ current = actual impedance, compare with rated impedance - Impedance deviation ≤10% - Component voltage, current, and rated impedance;

[0137] Power flow rationality - power transmission principle - under non-generating conditions, power flow is from the grid to the load - no reverse power (±1% fluctuation allowed) - active power, reactive power;

[0138] Based on the trigger event type and device ID of the high-confidence multi-parameter fusion dataset, the corresponding audit rules in the system-level physical law audit rule library are matched: if it is an electrical trigger event (such as voltage surge, current over-limit), the system power conservation, impedance matching verification and distribution node current balance rules are loaded first; if it is a mechanical trigger event (such as circuit breaker tripping), the power flow rationality rule is loaded additionally (to verify whether the load power returns to zero after tripping); the rated parameters (such as rated impedance, rated power, etc.) in the high-confidence multi-parameter fusion dataset are automatically extracted, and the rule verification threshold is configured without manual intervention;

[0139] It should be noted that a distribution node refers to a unit in the shore power system that undertakes the function of power distribution (such as a berth distribution cabinet, busbar tap point, transformer output node, etc.). A distribution node will deploy multiple monitoring nodes to achieve full parameter coverage and collection. For example, a berth distribution node (functional unit) may deploy 3 current monitoring nodes (collecting three-phase current), 1 voltage monitoring node (collecting busbar voltage), and 1 temperature monitoring node (collecting temperature rise inside the cabinet). During auditing, the fused data of all monitoring nodes under this distribution node will be aggregated, and audit verifications such as current balance and power conservation will be performed.

[0140] Based on the matching audit rules, corresponding audits are performed according to the audit dimensions to identify data anomalies or equipment anomalies; the audits include current balance audits of distribution nodes, system power conservation audits, impedance matching and power flow audits;

[0141] Among them, the current balance audit of distribution nodes involves: extracting the three-phase current (I_A, I_B, I_C) and zero-sequence current (I_0) data of each distribution node at the same timestamp based on a high-reliability multi-parameter fusion dataset; and calculating the current balance degree as |I_A+I_B+I_C-I_0|÷Max(I_A,I_B,I_C)×100%.

[0142] The results are judged based on the calculation results. A current balance of ≤5% is considered qualified; 5% < current balance ≤8% is considered slightly abnormal; and current balance >8% is considered seriously abnormal. The abnormal node ID, timestamp, and specific current value are marked.

[0143] System power conservation audit: Extract the total input power (P_in), total output power (P_out), and system loss power (P_loss) of the entire power distribution system at the same timestamp.

[0144] Calculate power deviation = |P_in - (P_out + P_loss)| ÷ P_in × 100%;

[0145] Result judgment: power deviation ≤ 3% is qualified; 3% < power deviation ≤ 5% is a minor abnormality; power deviation > 5% is a serious abnormality. Mark the power deviation value and the corresponding timestamp.

[0146] Impedance matching and power flow auditing:

[0147] Impedance matching: Extract voltage and current data of key components at the same time stamp, calculate the actual impedance (i.e., the ratio of voltage to current), and compare it with the rated impedance. If the deviation is ≤10%, it is qualified; if it exceeds the limit, mark the component ID and abnormal impedance value.

[0148] Power flow direction: Extract active power data. If reverse power occurs under non-power generation conditions (P_in < 0 and duration ≥ 100ms), it is judged as an abnormal power flow direction, and the abnormal period and power value are marked.

[0149] Audits are performed point by point according to timestamps to ensure no audit blind spots; for slightly abnormal data, three timestamps need to be monitored continuously. If the abnormality continues, it is upgraded to a serious abnormality to avoid misjudgment caused by momentary interference.

[0150] Based on the audit results and combined with the genealogical labels of the high-confidence multi-electrical parameter fusion dataset, the source of the anomaly is located: node-level location (identifying the power distribution node and monitoring node to which the anomaly belongs), device-level location (locating the specific device associated with the anomaly), and time-level location (accurately marking the start and end timestamps of the anomaly).

[0151] By accessing full-process traceability data, the system traces and analyzes the causes of anomalies: at the data level, it checks the sensor health and fusion deviation corresponding to the anomaly data to determine if the problem is due to data acquisition exceeding the fusion threshold; at the environmental level, it correlates environmental data snapshots to analyze whether the equipment parameter drift is caused by harsh environments such as salt spray or vibration; at the equipment level, it combines historical equipment operating data to determine whether the anomaly is caused by gradual degradation.

[0152] Anomalies are categorized into three levels based on severity, and each level is labeled accordingly. The level classification is as follows: Level 1 (Severe): Balance > 8%, Power Deviation > 5%, Impedance Deviation > 10%, marked as requiring urgent handling; Level 2 (Minor): 5% < Balance ≤ 8%, 3% < Power Deviation ≤ 5%, marked as requiring monitoring; Level 3 (Potential): Single minor anomaly (not lasting for 3 timestamps), marked as requiring attention. Finally, all abnormal equipment components are integrated to form an abnormal equipment component list.

[0153] Integrate basic audit information (anchor ID, trigger event type, audit time range, list of nodes and devices involved in the audit), audit result statistics (percentage of qualified and abnormal audits in each dimension, distribution of abnormal levels, core abnormal nodes or devices), and list of abnormal devices and components to generate a structured audit report.

[0154] The methods for generating system-level physical audit reports include:

[0155] Based on the list of abnormal equipment components, multi-dimensional parameters of abnormal equipment components are extracted from a high-reliability multi-electrical parameter fusion dataset. These parameters include electrical characteristic parameters (impedance value, three-phase current, voltage fluctuation value, power loss, power factor, etc.), physical state parameters (component temperature rise, vibration acceleration, insulation resistance), and related parameters (environmental data corresponding to the collection period, and historical health data of the equipment). The extraction rules are based on a three-dimensional index of equipment component ID-parameter type-timestamp to ensure that the parameter time series completely overlaps with the audit anomaly period.

[0156] The extracted parameters of different dimensions were mapped to the [0, 1] interval through a normalization method to eliminate the influence of dimensional differences on the degradation trend analysis; then, low-confidence parameter data with health < 0.5 were removed, and the effective core parameter sequence was retained.

[0157] For the retained component multi-parameter data, a degradation feature correlation determination is performed. The core logic is as follows:

[0158] Set parameter degradation thresholds (e.g., impedance exceeding the rated value by 10%, temperature rise exceeding the industry standard by 20℃), and mark the degradation start time of a single parameter; then calculate the degradation time difference of multiple parameters. If the degradation start time difference of two or more parameters is ≤100ms, it is judged as a synergistic degradation characteristic; establish a correlation model, calculate the correlation between parameters using the Pearson correlation coefficient, and when the correlation coefficient is ≥0.7, it is judged as a strongly correlated degradation parameter pair (e.g., temperature rise-increase impedance, vibration aggravation-current fluctuation).

[0159] Based on the correlation analysis results, collaborative degradation trajectory curves of equipment components are generated. The trajectory generation rule is to draw a single trajectory for a single component. The trajectories of multiple components under the same power distribution node are summarized into a node-level collaborative degradation trajectory map.

[0160] The trajectory's horizontal axis is a time series based on an absolute spatiotemporal anchor point (covering 10ms before and after the anomaly), while the trajectory's vertical axis is a normalized multi-parameter value. The trajectory is marked with the collaborative degradation start time point, degradation rate mutation point, and degradation saturation point, and associated with the corresponding environmental conditions and audit anomaly types at each time point.

[0161] With collaborative degradation trajectory as the core, the system integrates audit results and full-process traceability data to construct a system-level physical audit report, which includes a basic information section (anchor ID, audit equipment range, parameter extraction period, data quality verification conclusion), a degradation trajectory analysis section (single component collaborative degradation trajectory curve, node-level trajectory map, strongly correlated degradation parameter pairs, degradation feature summary), an audit anomaly correlation section (audit anomaly type, anomaly location information, degradation trajectory and anomaly cause analysis), and a traceability link section (data acquisition-fusion-audit full-process log index, parameter data snapshot).

[0162] The generated collaborative degradation trajectories are categorized and archived according to equipment model and degradation type, and included in the equipment degradation case library. When auditing similar equipment in the future, the equipment degradation case library can be called to predict degradation trends, realizing the extension from post-abnormal audit to pre-abnormal warning.

[0163] The methods for recommending repair solutions by matching historical fault case databases include:

[0164] Based on the system-level physical audit report, multi-dimensional fault characteristics are extracted according to the three-level architecture of equipment, fault and environment tracing;

[0165] The equipment dimension (basic characteristics) includes equipment model, component ID, years of operation, installation berth, and current operating load; the fault dimension (core characteristics) includes fault type (data anomaly, equipment anomaly), anomaly level, co-deterioration parameter pairs, deterioration rate (parameter change per unit time), and co-deterioration start time difference; the environmental tracing dimension (related characteristics) includes fault triggering environmental conditions, full-process data quality score, and historical fault records (whether they have occurred repeatedly in the last 90 days).

[0166] To improve matching accuracy, quantization encoding is performed on the extracted unstructured and semi-structured features:

[0167] Numerical features (such as degradation rate and years of operation) are normalized to the [0,1] interval using the min-max method to eliminate dimensional differences.

[0168] Classification features (such as fault type, anomaly level): converted into numerical vectors through one-hot encoding;

[0169] Textual features (such as co-degraded parameter pairs and environmental conditions) are converted into fixed-dimensional feature vectors using word embedding (Word2Vec).

[0170] The final result is a standardized fault feature vector with dimension N×1 (N is the total number of features, which can be expanded according to equipment type).

[0171] Based on the fault feature vector, invalid cases are first filtered out in the pre-set historical fault case library according to the precise matching strategy. Priority is given to matching the two key fields of equipment model and component ID. Only cases that are completely matched are retained to enter the subsequent steps. If there are no completely matched cases for new equipment models, the matching is relaxed to equipment type + component function (such as shore power transformer equipment, winding function component).

[0172] The historical fault case library has a pre-defined standardized structure, containing multiple historical fault cases. Each case includes a fault feature set (deterioration parameter combination, deterioration rate, anomaly level, collaborative deterioration start time difference, triggering environmental conditions), equipment information (equipment model, component ID, years of operation, installation berth), maintenance plan (core maintenance steps, required tools or spare parts, operating procedures, safety precautions, estimated working hours), implementation effect (fault elimination rate, post-maintenance runtime, retest audit results), and operation and maintenance feedback (common problems, optimization suggestions, applicable scenario limitations).

[0173] Then, using the fuzzy matching strategy, the similarity between the current fault feature vector and the historical case feature vector is calculated through the cosine similarity algorithm;

[0174] The formula is: Similarity = (Current Feature Vector × Historical Case Feature Vector) ÷ (||Current Feature Vector|| × ||Historical Case Feature Vector||); The similarity value ranges from [0,1], and the closer it is to 1, the higher the matching degree.

[0175] By setting three levels of similarity thresholds (e.g., using 85% and 60% as threshold ranges), cases are divided into high-matching cases (≥85%, with a maximum of three highest similarity cases retained), medium-matching cases (≥60% and <85%, with a maximum of five cases retained), and low-matching cases (<60%). High-matching and medium-matching cases are selected, and low-matching cases are marked as invalid and directly removed. If no high-matching cases are found after selection (only medium-matching cases exist), case suitability adjustment suggestions need to be added. If there are no matching cases (all similarity cases <60%), a temporary repair plan generation process is triggered, indicating that manual on-site inspection and confirmation are required, and a historical fault case database update application is triggered simultaneously.

[0176] The repair plan with the highest similarity among the highly matched cases is used as the benchmark repair plan. The optimization suggestions from the second and third cases are integrated into the benchmark repair plan (such as tool replacement, spare parts general solution, etc.). Then, the repair steps of all medium-matched cases are combined with the current equipment operating conditions and on-site maintenance resources (such as available tools, spare parts inventory) to generate an adaptation adjustment plan. Finally, a repair plan recommendation report is compiled, including a fault feature summary (extracting core fault information (equipment or component, fault type, degradation characteristics)), case matching details (list of matched cases (including similarity, case source), description of the matching process), recommended repair plan (benchmark plan (repair steps, tool or spare parts list, estimated working hours, safety precautions), alternative plan (only provided when highly matched cases are insufficient)) and implementation suggestions (adaptation adjustment points, key points of on-site inspection, post-repair retesting requirements).

[0177] If the current fault has no matching case and the fault is eliminated after the repair is implemented, the fault characteristics, repair plan and implementation effect are compiled into a new case and added to the historical fault case library; at the same time, the case library index is updated to improve the efficiency of subsequent matching. If the repair plan is not effective (the fault is not eliminated), the reason for the failure of the plan is recorded to provide a basis for the iteration of the historical fault case library.

[0178] The methods for generating operation and maintenance decision work orders include:

[0179] Based on the maintenance solution recommendation report and the system-level physical audit report, for multi-berth parallel operation scenarios, the current operating data of the multi-berth parallel equipment cluster is collected, including the load rate of each berth, the voltage and frequency of the parallel bus, the power flow direction, the status of the backup power supply, and historical load fluctuation data (last 24 hours); simultaneously, the equipment ledger of the multi-berth parallel equipment cluster (including berth number, equipment model, parallel topology relationship, load capacity limit), operation and maintenance resource configuration information (skill level of on-site operation and maintenance personnel, available tools or spare parts inventory, backup berth activation status), and safety specification data of industry or enterprise standards are obtained;

[0180] Then, assess the operating status and load distribution characteristics of the device cluster, specifically:

[0181] Focusing on the core risks of multi-berth parallel operation scenarios, we extract and analyze key parameters, including load distribution status, parallel operation coordination status, and fault impact range.

[0182] Among them, the load distribution status is as follows: the current load rate and load fluctuation coefficient of each berth are statistically analyzed to identify heavily loaded berths (load rate > 85%) and lightly loaded berths (load rate < 50%); the parallel operation coordination status is as follows: the voltage deviation, frequency deviation, and phase synchronization of the parallel operation bus are checked; the fault impact range is as follows: based on the parallel operation topology, it is determined whether the faulty berth affects the power supply of adjacent berths (such as bus connection, load linkage), and if so, it is marked as a high-risk associated berth;

[0183] Based on the operating status of the device cluster, predict the feasibility of refined operation and maintenance linkage;

[0184] The process begins with assessing the feasibility of load transfer: evaluating whether the remaining load capacity of lightly loaded or standby berths meets the load transfer requirements of the faulty berth (remaining capacity ≥ 1.2 times the transferred load); then confirming that the equipment cluster has the hardware to monitor the load transfer process in real time (such as current and voltage sensors, audible and visual alarm devices, etc.); finally verifying whether on-site maintenance personnel have the qualifications to operate the parallel equipment cluster system and whether the number of available personnel meets the requirements for both operation and monitoring positions; if all requirements are met, then the multi-berth parallel operation scenario is deemed feasible for linkage.

[0185] For multi-berth parallel operation scenarios with linkage feasibility, load transfer logic, process monitoring thresholds and safety interlocks are pre-set, and standardized coding is performed according to berth ID, operation type, parameter threshold and execution sequence to build operation and maintenance linkage instructions that can be directly issued.

[0186] The load transfer logic is designed based on the principles of safety first and minimal impact. The design dimensions include: determining the transfer target (prioritizing the transfer of non-core loads (such as auxiliary equipment) of the faulty berth, and then transferring core loads; the total transfer load = the current load of the faulty berth × 1.1 (with redundancy); if a voltage drop occurs during the load transfer process, the transfer is immediately suspended); selecting the transfer target (prioritizing lightly loaded berths on the same busbar (load rate < 50%), and then using backup berths; avoiding transfer to heavily loaded berths (load rate > 80%); if the load rate of the target berth suddenly rises to over 90%, switch to the backup target); controlling the transfer timing (adopting a stepped transfer: the load transferred in each step is ≤ 30% of the remaining capacity of the target berth, and the interval between steps is ≥ 30 seconds; the total transfer time is ≤ 10 minutes; if a single step transfer exceeds 5 minutes, manual intervention is triggered); and power switching logic (confirming the power supply of the target berth is stable before the transfer (voltage fluctuation ≤ ±1%), and implementing a first-close-then-disconnect strategy during switching (to avoid power outages); if the phase deviation exceeds 5° during the switching process, immediately disconnect the switching switch).

[0187] The process monitoring threshold is the monitoring threshold for the entire operation and maintenance linkage process, and is equipped with safety interlocking instructions to ensure that the operation and maintenance linkage is safe and controllable. The process monitoring threshold is designed as follows: during the load transfer process, the voltage deviation of the parallel bus is ≤ ±3% of the rated value, the frequency deviation is ≤ ±0.8Hz, and the current imbalance is ≤ 5%. After the fault berth is de-energized, the residual voltage is ≤ 10% of the rated value.

[0188] The safety interlock command is designed as follows: when the monitored parameter exceeds the process monitoring threshold, it will automatically trigger a pause transfer and an audible and visual alarm; if the limit is exceeded for 3 seconds, it will execute an emergency power cut-off + load unloading command; after the transfer is completed, it will trigger a target berth load stability monitoring command (lasting 5 minutes).

[0189] By combining the synergistic degradation trajectory with the fault characteristics and the parallel operation status of multiple berths, the root cause of the fault is inferred. For example, if the fault is a synergistic degradation of increased impedance and increased temperature, the root cause is inferred to be winding aging or poor contact. If a slight current imbalance occurs simultaneously in multiple berths, the root cause is inferred to be phase deviation of the parallel bus. If the fault occurs during periods of high salt spray or high humidity, the root cause of the component insulation degradation due to environmental corrosion is supplemented.

[0190] Integrate the operation and maintenance linkage instructions, the baseline maintenance plan and adaptability adjustment plan in the maintenance plan recommendation report, as well as the implementation requirements (including operation and maintenance personnel configuration, implementation time requirements, post-maintenance retesting standards, etc.) and the root cause of the fault to form an operation and maintenance decision work order;

[0191] It should be noted that the core application scenario of this solution is multi-berth parallel operation, but the solution also has a certain degree of scalability and is still applicable to non-parallel operation scenarios. It is only necessary to remove the multi-berth linkage command link and directly integrate the root cause and maintenance solution to generate operation and maintenance decision work orders.

[0192] The standard case archive is packaged and executed through a dual-path update mechanism to achieve closed-loop self-evolution, including:

[0193] Acquire data on the execution process of operation and maintenance decision work orders (operation records of operation and maintenance personnel, instruction issuance logs, on-site environmental data and resource consumption data), combine it with fault verification results and post-repair retest data (data of retest evaluation is re-collected according to the entire process of all preceding links), clean the acquired data and remove redundancy, and then classify the cleaned data according to the three-dimensional structure of work order ID-data type-link, and establish cross-link data association mapping;

[0194] Among them, Category Dimension 1 (Work Order ID): uses the unique identifier of the operation and maintenance decision work order as an index to bind all data corresponding to the work order;

[0195] Category Dimension 2 (Data Type): Divided into three main categories: execution process, verification result, and retest.

[0196] Category Dimension 3 (Steps): Corresponds to the preceding full-process steps, such as the data association audit step for retesting;

[0197] Association rules are defined as follows: ensure the causal correspondence between fault verification results, repair plans, and retest data, such as the root cause verification conclusions needing to be consistent with the fault elimination conclusions in the retest audit report;

[0198] The classified and associated data is integrated with the preceding full-process data, packaged into standard cases and stored in the historical failure case library. If the similarity with existing cases in the library is ≥90%, it is determined to be a similar case, and the execution effect data and retest conclusions of the existing case are updated. If the similarity is <90%, it is determined to be a new case and a unique case number is assigned.

[0199] Based on the cases entered into the database, the evaluation system is updated through a dual-path update mechanism that optimizes auditing and matching rules, as well as adjusting health parameters. Specifically:

[0200] The audit and matching rules have been optimized as follows: The accuracy and false positive rates of physical audit rules for fault identification, as well as the success rate and adaptability of the recommended solutions for case matching rules, have been statistically analyzed. For rules with high false positive rates or high adaptability, the verification thresholds have been adjusted based on actual test data from the cases. For example, if the original acceptable threshold for current balance was 5%, and a normal fluctuation of 4.8% was repeatedly judged as abnormal, the threshold has been optimized to 6%. A distinction has been made between multi-berth parallel operation scenarios and single-berth independent scenarios, and priorities have been set for rules in different scenarios. In multi-berth scenarios, rules related to load transfer and parallel operation coordination have higher priority than general rules. In single-berth scenarios, linkage-related rules have been simplified, focusing on equipment-specific fault audit rules.

[0201] The health parameters are adjusted as follows: based on the equipment degradation trajectory of the associated standard case and the health retest data after maintenance, the health weights and threshold parameters are adjusted.

[0202] Weight parameter adjustment: If a certain parameter (such as temperature rise) is a core precursor to deterioration, increase the weight of that parameter in the health calculation (e.g., increase the physical weight from 15% to 18%, and decrease other weights accordingly).

[0203] Threshold parameter adjustment: If the health threshold of a certain type of device is too low, causing the health labeling to lag, adjust the health classification threshold of the device (e.g., adjust the health threshold from 0.6 to 0.7).

[0204] For the evaluation system updated by the dual-path update mechanism, effectiveness verification is performed. The verification method is to select 3 to 5 similar cases and perform a simulated full process of data collection, analysis, decision-making, and execution. If the fault identification rate improves (e.g., ≥5%), the case matching accuracy improves (e.g., ≥8%), and the timeliness of health warnings improves (e.g., ≥10%), the verification is deemed successful, and the updated evaluation system officially takes effect, covering the original system. If the standards are not met, the dual-path update is performed again. Finally, a complete closed-loop evaluation management link of "data collection, analysis, decision-making, execution, and optimization" is formed. Example 2:

[0205] Please see Figure 3 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. A condition assessment system for shore power transfer devices based on multi-parameter measurement is provided, including:

[0206] Synchronous acquisition and slice generation module: Trigger evaluation through device status signals, generate and broadcast absolute spatiotemporal anchor points, drive all monitoring nodes to sample synchronously; At the same time, combine historical device health and environmental data to dynamically adjust the periodic sampling frequency, capture multi-electrical parameter data, and integrate to generate multi-electrical parameter process slice dataset;

[0207] Health assessment and data fusion module: Based on the multi-electrical parameter process slice dataset, the dynamic health of the data is calculated according to the progressive assessment rules to form an integrated data sequence; through three-level filtering and weighting, a high-confidence multi-electrical parameter fusion dataset with complete lineage labels is generated;

[0208] Physical Audit and Degradation Trajectory Module: Performs system-level physical law audit on high-confidence datasets, locates fault sources, generates a list of abnormal equipment components, extracts multiple parameters of equipment components to generate collaborative degradation trajectories, and integrates them to form a system-level physical audit report;

[0209] Fault matching and maintenance decision module: Extract fault characteristics based on physical audit reports, match and recommend maintenance solutions from historical fault case database, and generate maintenance decision work orders with maintenance linkage instructions for multi-berth parallel operation scenarios;

[0210] Data archiving and system optimization module: Based on the execution results of operation and maintenance decision work orders, standard case archives are packaged and formed. A closed-loop self-evolution is performed through a dual-path update mechanism to form an evaluation and management closed-loop link. Example 3:

[0211] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the aforementioned shore power transfer device condition assessment system based on multi-electrical parameter measurement.

[0212] Since the electronic device described in this embodiment is the electronic device used to implement the shore power transfer device condition assessment method based on multi-electrical parameter measurement in the embodiments of this application, those skilled in the art can understand the specific implementation and various variations of the electronic device in this embodiment based on the shore power transfer device condition assessment method based on multi-electrical parameter measurement described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the shore power transfer device condition assessment method based on multi-electrical parameter measurement in the embodiments of this application falls within the scope of protection of this application.

[0213] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0214] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for assessing the condition of a shore power transfer device based on multi-parameter measurement, characterized in that, include: S1: Trigger evaluation through device status signals, generate and broadcast absolute spatiotemporal anchor points, and drive all monitoring nodes to sample synchronously; Simultaneously, the periodic sampling frequency is dynamically adjusted by combining the equipment's historical health status and environmental data to capture multi-electrical parameter data and integrate them to generate a multi-electrical parameter process slice dataset. The methods for generating the absolute spatiotemporal anchor point include: Continuously capture mechanical status signals and electrical operation signals of key component monitoring nodes of the shore power unit at the berth, use them as equipment status signals, and decode and map them into equipment operating status; The device's operating status, after real-time decoding and mapping, is compared and determined with a pre-set trigger judgment rule base. Once the trigger evaluation conditions are met, an absolute spatiotemporal anchor point containing a unique identifier and a precise timestamp is generated based on the standard time and broadcast to all monitoring nodes via high-priority communication. S2: Based on the multi-parameter process slice dataset, the dynamic health of the data is calculated according to the progressive evaluation rules to form an integrated data sequence; through three-level filtering and weighting, a high-confidence multi-parameter fusion dataset with complete lineage labels is generated; The integrated data sequence is formed in the following ways: Based on the multi-parameter process slice dataset, the health of each data point is comprehensively calculated according to the progressive evaluation rules of sensor self-test, environmental dynamic threshold adjustment, redundancy reading consistency verification and physical limit paradox check, combined with weight allocation. The health level label and related information are attached to the data points to form an integrated data sequence. S3: Perform system-level physical law audit on high-confidence datasets, locate fault sources, generate a list of abnormal equipment components; extract multi-parameters of equipment components to generate collaborative degradation trajectories, and integrate them to form a system-level physical audit report; Based on a high-reliability multi-electrical parameter fusion dataset, the system matches corresponding audit rules through a system-level physical law audit rule base, and performs matching power distribution node current balance audit, system power conservation audit, impedance matching and power flow audit by audit dimension to identify data anomalies or equipment anomalies. The system-level physical audit report is generated in the following ways: Based on the list of abnormal equipment components, multi-dimensional parameters of abnormal equipment components are extracted from a high-confidence multi-electrical parameter fusion dataset, and the starting time point of single parameter degradation is marked. Furthermore, through multi-parameter degradation time difference analysis and correlation modeling, the characteristics of collaborative degradation are determined, and the collaborative degradation trajectory of equipment components is generated; By integrating audit results and full-process traceability data with collaborative degradation trajectory as the core, a system-level physical audit report is constructed. S4: Extract fault characteristics based on physical audit reports, match and recommend maintenance solutions from the historical fault case library, and generate maintenance decision work orders containing maintenance linkage instructions for multi-berth parallel operation scenarios; S5: Based on the execution results of operation and maintenance decision work orders, package them into standard case archives, execute closed-loop self-evolution through a dual-path update mechanism, and form an evaluation and management closed-loop link; The packaging process forms a standard case archive, and the closed-loop self-evolution is performed through a dual-path update mechanism, including: Obtain data on the execution process of operation and maintenance decision work orders, combine it with fault verification results and post-repair retest data, perform structured classification and association, package it into standard cases and store it in the historical fault case library; Based on the cases entered into the database, the evaluation system is updated through a dual-path update mechanism that combines auditing and matching rule optimization with health parameter adjustment, and validity verification is performed. The system takes effect after the verification is successful.

2. The method for assessing the condition of a shore power transfer device based on multi-parameter measurement according to claim 1, characterized in that, The generation methods for the multi-parameter process slice dataset include: Each monitoring node synchronously starts sampling based on an absolute spatiotemporal anchor point, and simultaneously combines historical equipment health data with environmental data to construct a two-dimensional judgment matrix, dynamically adjusting the periodic sampling frequency of core components; then, using the adjusted sampling frequency, it captures full-cycle multi-electrical parameter data of the triggering event. The collected multi-electrical parameter data is aligned with anchor reference time and filtered for validity. The filtered valid data is then integrated into a structured dataset based on time series and parameter type to form a multi-electrical parameter process slice dataset.

3. The method for assessing the condition of a shore power transfer device based on multi-parameter measurement according to claim 2, characterized in that, The methods for generating the high-reliability multi-parameter fusion dataset include: Based on the integrated data sequence, the raw multi-electrical parameter data of each monitoring node are filtered in three levels according to the health level label to retain the effective core data; The effective core data are weighted and fused with health status as the weight, and a complete genealogical label is added to the fused data to form a highly reliable multi-electrical parameter fusion dataset.

4. The method for assessing the condition of a shore power transfer device based on multi-parameter measurement according to claim 1, characterized in that, The methods for generating the list of abnormal equipment components include: For any abnormal results discovered during the audit, we will locate and trace the source, analyze the causes by combining the data from the entire process of traceability, mark the level of abnormality, and integrate and generate a list of abnormal equipment and components.

5. The method for assessing the condition of a shore power transfer device based on multi-parameter measurement according to claim 1, characterized in that, The methods for recommending repair solutions by matching historical fault case databases include: Based on the system-level physical audit report, multi-dimensional fault features are extracted according to the three-level architecture of equipment, fault and environment tracing, and standardized fault feature vectors are generated through quantitative coding processing. Based on the fault feature vector, invalid cases are filtered out in the pre-set historical fault case library using an exact matching strategy, and the similarity between the current and historical case feature vectors is calculated using a fuzzy matching strategy to select high-matching and medium-matching cases. Based on the repair solution with the highest similarity among the highly matched cases, and taking into account the repair steps, current operating conditions and maintenance resources of the medium-matched cases, an adaptation adjustment plan is generated and integrated into a repair solution recommendation report.

6. The method for assessing the condition of a shore power transfer device based on multi-parameter measurement according to claim 5, characterized in that, The methods for generating the operation and maintenance decision work order include: Based on the maintenance plan recommendation report and the system-level physical audit report, for multi-berth parallel operation scenarios, the current operation data of the multi-berth parallel operation equipment cluster is collected and correlated to assess the operation status and load distribution characteristics of the equipment cluster and predict the feasibility of operation and maintenance linkage. For multi-berth parallel operation scenarios with linkage feasibility, operation and maintenance linkage instructions are constructed through pre-set load transfer logic, process monitoring thresholds and safety interlocks, and maintenance plans are associated to form operation and maintenance decision work orders.

Images