Marine cable on-line fault judgment automatic detection method

Abstract

The invention relates to the technical field of electrical fault detection, in particular to an online fault judgment and automatic detection method for a marine cable, which comprises the following steps of: acquiring ground leakage current and load current in operation, extracting an active component of the leakage current and analyzing the nonlinear correlation between the active component and the load current to realize online early warning of insulation degradation; when the ship is in a working condition of allowing short-time power failure of partial non-critical loops, positioning a fault branch based on a normalization result of the ground capacitance variation before and after the power failure of each branch and the cable length; after power supply is recovered, the fault type is judged by combining the leakage current fundamental wave and phase voltage phase difference and the harmonic distortion rate; during power-off maintenance, high-frequency carrier waves are injected, the position of a fault point is determined according to current attenuation abnormity, correction is carried out by combining temperature and vibration factors, and the detection accuracy and the positioning precision are improved.

Description

Technical Field

[0001] This invention relates to the field of electrical fault detection technology, specifically to an automatic online fault detection method for marine cables. Background Technology

[0002] Marine electrical systems are isolated power grids, and to ensure power continuity, they generally employ IT grounding, meaning the neutral point is either ungrounded or grounded with high resistance. This grounding method is characterized by the system not tripping upon the first insulation fault and continuing operation; however, a second grounding fault could lead to a complete power outage, jeopardizing navigational safety. Therefore, effective fault detection of marine cables is of great importance.

[0003] Currently, cable fault detection technologies mainly include time-domain reflectometry, frequency-domain reflectometry, insulation resistance measurement, and online electrical parameter monitoring. However, all of these technologies have inherent limitations in marine applications. Power cables for critical ship systems such as steering gear, main engine control, and navigation equipment cannot be tested while the power is off, making high-precision offline detection methods unusable. More critically, when an insulation fault occurs in an IT grounding system, the system does not trip and the load continues to operate normally, so electrical parameters such as current and voltage do not show abnormalities. This causes online monitoring methods based on operational anomaly detection principles to completely fail in the early stages of a fault. Furthermore, marine cables are energized from the date of installation, making it impossible to establish reliable health baseline parameters through power-off measurements. This leaves existing fault judgment methods based on threshold comparisons lacking effective comparative criteria. Therefore, a detection method is needed that can achieve automatic online identification and precise location of cable faults from normal operating signals without power interruption and without abnormal electrical parameter characteristics.

[0004] Therefore, an automatic detection method for online fault diagnosis of marine cables is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic detection method for online fault diagnosis of marine cables. By analyzing the correlation between the active component of leakage current and load current, it can realize early warning of insulation degradation. Under specific operating conditions, it can locate faulty branches based on changes in capacitance to ground. Furthermore, it can combine phase difference, harmonic characteristics, and high-frequency carrier attenuation to achieve fault type identification and accurate location.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An automatic detection method for online fault diagnosis of marine cables includes: The system collects the ground leakage current and load current during operation, extracts the active component of the leakage current, analyzes the correlation between the active component and the load current, and determines that there is an insulation degradation fault when the active component is nonlinearly positively correlated with the load current. When the ship is in a condition where short-term power outages of some non-critical circuits are permitted, the capacitance to ground of each power-disconnectable branch is obtained before and after sequential disconnection. The change in capacitance to ground of each branch is calculated, and the change in capacitance is normalized with the corresponding branch cable length. Branches whose normalized values ​​exceed the normal range are identified as faulty branches. After power is restored to each branch, the leakage current to ground and phase voltage are collected for the faulty branch. The phase difference between the fundamental component of the leakage current and the voltage and the harmonic distortion rate are extracted. Based on the combined characteristics of the phase difference and the harmonic distortion rate, the fault type is determined to be insulation deterioration, insulation breakdown and / or intermittent fault. When the faulty branch is de-energized for maintenance, a high-frequency carrier signal is injected into the cable, and the high-frequency current amplitude at multiple accessible locations along the cable is measured. The fault point is determined based on the location segment where the high-frequency current attenuation rate between adjacent measuring points increases abnormally.

[0007] Preferably, extracting the active component of the leakage current includes: performing spectral decomposition on the collected leakage current to obtain the fundamental component, extracting the component of the fundamental component that is in phase with the phase voltage as the active component, and extracting the component of the fundamental component that is orthogonal to the phase voltage as the reactive component. The analysis of the correlation between active power components and load current includes: synchronously recording time series data of active power components and load current within a continuous time window, calculating the rate of change of active power components with respect to load current, and determining that when the rate of change is greater than a first preset threshold and maintains a monotonically increasing trend within multiple continuous time windows, it is a nonlinear positive correlation.

[0008] Preferably, the ship is determined to be in a condition where a short-term power outage of some non-critical circuits is permitted by the following method: obtaining information on the ship's main engine speed, propeller speed, generator load rate, and ship position; when the main engine speed is lower than a set percentage of the rated speed, the propeller speed is zero, and the ship position is within the port coordinate range, the ship is determined to be in a berthing condition; and / or when the main engine speed is zero, the generator load rate is lower than a second set threshold, and the ship position remains stationary for more than a set duration, the ship is determined to be in an anchoring condition. The normal range is determined as follows: extract the ratio of the change in capacitance to ground to the length of cables of the same type and specification, remove outliers that deviate from the mean by more than a set multiple, calculate the mean and standard deviation of the remaining ratios, and take the interval between the mean and three times the standard deviation as the normal range.

[0009] Preferably, the fault type is determined based on the combined characteristics of phase difference and harmonic distortion rate, including: when the phase difference is between 90 degrees capacitive lead and a set first angle capacitive lead and the harmonic distortion rate is lower than a set first threshold, it is determined to be mild insulation degradation. When the phase difference is between the capacitive lead setting first angle and the capacitive lead setting second angle and the harmonic distortion rate is lower than the set second threshold, it is judged as moderate insulation degradation. When the phase difference is between the capacitive lead setting second angle and the in-phase and the harmonic distortion rate is higher than the setting second threshold, it is judged as insulation breakdown; When the leakage current exhibits intermittent pulse changes and the frequency of these changes exceeds a set frequency threshold, it is identified as an intermittent fault.

[0010] Preferably, the frequency of the high-frequency carrier signal is determined based on the cable length and the expected positioning accuracy: the total cable length of the faulty branch is obtained, the minimum frequency required is calculated based on the propagation speed of the high-frequency signal in the cable and the required positioning accuracy, and a frequency that avoids the frequency band of ship power interference is selected as the carrier frequency based on the minimum frequency. The selection of multiple accessible locations along the cable includes setting measurement positions at the cable head, branch junction box locations, cable tray seals, cable support fixing points, and cable ends. The interval between measurement positions is determined according to the required positioning accuracy.

[0011] Preferably, after determining that an insulation degradation fault exists, the amplitude change trend of the neutral point to ground voltage is continuously monitored, and the change trend of the neutral point to ground voltage and the change trend of the active component of the leakage current are analyzed in a time series correlation. When the neutral point to ground voltage increases monotonically and shows a synchronous growth trend with the active component of the leakage current, it is used as an auxiliary confirmation basis for the insulation degradation fault.

[0012] Preferably, ambient temperature data is collected at multiple locations along the cable, a statistical relationship model between the active component of leakage current and ambient temperature is established, the expected leakage current value based on temperature influence is calculated according to the current ambient temperature, the actual measured active component of leakage current is compared with the expected value, and when the actual value exceeds the set multiple of the expected value, the excess part is determined to be abnormal leakage caused by insulation deterioration. The vibration acceleration signal of the ship's main engine is collected, the main frequency and amplitude of the vibration are extracted, and the time-domain fluctuation of leakage current and / or ground capacitance is analyzed with the vibration signal in the frequency domain. When the fluctuation frequency of leakage current and / or ground capacitance is found to be consistent with the main frequency of vibration and the correlation coefficient exceeds the set threshold, it is determined to be an intermittent fault induced by vibration.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves online identification of insulation degradation status by extracting the active component of the leakage current to ground and analyzing its nonlinear positive correlation with the load current. The invention introduces a rate of change and continuous time window monotonicity analysis mechanism, which can distinguish between normal load fluctuations and abnormal growth trends caused by degradation, significantly reducing the false alarm rate. Simultaneously, by separating the active and reactive components, the interference of purely capacitive leakage is weakened, making the detection results closer to the actual aging mechanism of the insulation material. This allows for early warning before insulation breakdown, improving the operational safety of the ship's power system and the initiative in maintenance planning.

[0014] 2. This invention, under permissible short-term power outage conditions on ships, acquires the changes in ground capacitance of each branch before and after the power outage, normalizes this data with cable length, and establishes a unified discrimination standard. This scheme constructs normal ranges through statistical methods, eliminates outliers, and identifies abnormal branches, reducing reliance on human experience. Simultaneously, it automatically determines the detection timing based on ship operating status parameters, avoiding impact on critical circuits and balancing safety and practicality. This method is suitable for complex multi-branch power distribution network environments, enabling rapid identification of problem areas without large-scale power outages, significantly shortening troubleshooting time.

[0015] 3. After power is restored to a faulty branch, this invention analyzes the combined characteristics of the phase difference between the fundamental frequency of the leakage current and the phase voltage, as well as the harmonic distortion rate, to classify and differentiate between mild degradation, moderate degradation, insulation breakdown, and intermittent faults, achieving a deeper diagnosis from "whether there is a fault" to "what kind of fault." Furthermore, by injecting a high-frequency carrier signal during maintenance and power outage, the fault location is determined by measuring abnormal sections of high-frequency current attenuation rate at multiple measurement points along the line, achieving quantitative location. Combined with a temperature compensation model and vibration frequency domain correlation analysis, the influence of environmental and mechanical disturbances can be eliminated, improving the reliability of the judgment. Overall, a closed-loop detection system is formed, encompassing online monitoring, branch location, type identification, and precise point location. Attached Figure Description

[0016] Figure 1 This invention provides a schematic flowchart of an automatic online fault diagnosis and detection method for marine cables. Figure 2 This is a schematic diagram of the insulation degradation nonlinear correlation determination process provided by the present invention; Figure 3 This is a schematic diagram of the vibration-induced intermittent fault determination process provided by the present invention. Detailed Implementation

[0017] ZW0126010893 To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0018] This invention provides an automatic online fault detection method for marine cables, applicable to marine power systems employing IT grounding systems. This method collects operational electrical parameters and utilizes the physical characteristics of the IT system to achieve early identification of insulation degradation without interrupting power supply, and completes fault location and type determination under permissible power outage conditions.

[0019] The data acquisition point of the detection system is located at the ship's main switchboard. It collects ground leakage current through a zero-sequence current transformer and load current and phase voltage through current and voltage transformers. Data acquisition employs a synchronous sampling mechanism to ensure that the time deviation of each channel signal is less than 0.5 milliseconds, and the sampling frequency is set to above 2000 Hz. The acquired raw data is converted from analog to digital before entering the data processing unit for analysis.

[0020] Example 1: Please see Figure 1 This invention provides an automatic online fault detection method for marine cables. The technical solution is as follows: Collect the ground leakage current and load current during operation; extract the active component of the leakage current; analyze the correlation between the active component and the load current; when the active component shows a non-linear positive correlation with the load current, an insulation degradation fault is determined; when the ship is in a condition where short-term power outages of some non-critical circuits are permissible, obtain the ground capacitance values ​​before and after sequential disconnection of each disconnectable power distribution branch; calculate the change in ground capacitance of each branch; normalize the capacitance change with the corresponding branch cable length; and then... Branches with normalized values ​​exceeding the normal range are identified as faulty branches. After power is restored to each branch, leakage current to ground and phase voltage are collected for the identified faulty branches. The phase difference between the fundamental component of the leakage current and the voltage, as well as the harmonic distortion rate, are extracted. Based on the combined characteristics of the phase difference and harmonic distortion rate, the fault type is determined to be insulation degradation, insulation breakdown, and / or intermittent fault. During power outage maintenance of the faulty branch, a high-frequency carrier signal is injected into the cable, and the high-frequency current amplitude at multiple accessible locations along the cable is measured. The fault point is determined based on the location segment where the high-frequency current attenuation rate between adjacent measuring points increases abnormally.

[0021] During normal operation of the ship's electrical system, leakage current to ground is continuously collected using zero-sequence current transformers installed at the main switchboard busbar. Even without a ground fault, a small leakage current to ground will still occur in the IT system due to the distributed capacitance of the cables to ground. The three-phase load current of this distribution circuit is simultaneously collected and measured using current transformers installed on each phase conductor.

[0022] The acquired time-domain waveform of the ground leakage current is subjected to spectral decomposition. A Fast Fourier Transform (FFT) is used to convert the time-domain signal to a frequency-domain signal, extracting the fundamental component at 50 Hz. The fundamental component contains amplitude and phase information. The system phase voltage is measured synchronously, and the amplitude and phase angle of the fundamental component of the phase voltage are extracted. The fundamental component of the leakage current is projected onto the direction of the fundamental component of the phase voltage to obtain the component in phase with the voltage; this component is the active component of the leakage current, corresponding to resistive leakage current. The fundamental component of the leakage current is projected onto a direction orthogonal to the phase voltage to obtain the component that leads or lags the voltage by 90 degrees; this component is the reactive component of the leakage current, corresponding to capacitive or inductive leakage current.

[0023] The active component of leakage current and the values ​​of load current are recorded synchronously within a continuous time window. The length of the time window is set to 10 to 30 minutes to ensure that it can cover the typical fluctuation cycle of the load current. The judgment threshold is determined based on the health baseline value at the time of ship delivery. When the measured rate of change reaches 3 times the baseline value and increases monotonically over three consecutive windows, nonlinear degradation caused by thermoelectric coupling effect is confirmed.

[0024] The rate of change of the active power component with respect to the load current is calculated as the increment of the active power component divided by the increment of the load current. Under normal circumstances, the cable insulation resistance is extremely high, the active power leakage current is minimal and essentially constant, and the rate of change is close to zero. When insulation degradation occurs, the insulation resistance decreases, the load current increases, leading to an increase in the cable conductor temperature. Heat is conducted to the insulation layer, further increasing the insulation temperature. Since the resistivity of the insulation material has a negative temperature coefficient, the increased temperature causes a further decrease in insulation resistance, resulting in a corresponding increase in the active power leakage current. This thermoelectric coupling effect causes a nonlinear positive correlation between the active power component and the load current.

[0025] Determination of the rate of change threshold: During the mooring trials or first voyage before ship delivery, benchmark tests are conducted on all power distribution circuits under different load conditions. Active component leakage current and load current data are collected under no-load, half-load, and full-load conditions, and the rate of change of the active component with respect to the load current is calculated. The maximum rate of change measured under the three conditions is recorded as the benchmark rate of change. Three times the benchmark rate of change is taken as the judgment threshold and stored in the detection system. This threshold is within the range of two to five times the benchmark value.

[0026] When the rate of change measured in real time exceeds the threshold, an anomaly is initially identified. Further observation of the rate of change trend is conducted over multiple consecutive time windows. If the rate of change is greater than the threshold and exhibits a monotonically increasing trend over three consecutive time windows, it is confirmed that the active power component and load current show a non-linear positive correlation, indicating an insulation degradation fault. The specific procedure is as follows: Figure 2 The more windows there are, the higher the reliability of the judgment, but the longer the response time.

[0027] There are operating conditions during ship operation where some non-critical circuits can be temporarily disconnected, mainly including berthing and anchoring conditions.

[0028] Determining berthing conditions: First, acquire the main engine speed signal provided by the ship's automation system, which is measured by a speed sensor mounted on the main engine crankshaft. Compare the real-time measured main engine speed with the pre-stored rated speed. The first condition is met when the main engine speed is lower than 20% of the rated speed. This percentage threshold is determined based on the fact that the main engine is usually in an idling or low-speed warm-up state when the ship is berthing.

[0029] Second, the propeller speed sensor signal is acquired; this sensor is mounted on the propeller shaft. The second condition is met when the propeller speed is zero. In practice, sensor measurement error is considered; a speed less than five revolutions per minute is considered zero.

[0030] Third, obtain the ship's position coordinates from its Global Positioning System (GPS), including longitude and latitude data. The system pre-stores the coordinate ranges of each port, defined as a rectangular area defined by the port's center coordinates plus or minus 0.05 degrees. Compare the current ship's coordinates with all the pre-stored port coordinate ranges; the third condition is met when the ship's position falls within the longitude and latitude range of any port.

[0031] When all three conditions are met, the vessel is determined to be in berthing condition.

[0032] Determining the anchoring condition: First, the main engine speed is zero; less than five revolutions per minute is considered zero.

[0033] Second, acquire the generator power meter signal, which is converted into an electrical signal by a power transmitter. Calculate the ratio of the actual generator output power to the rated power to obtain the load rate. The second condition is met when the generator load rate is below 30%. This threshold is determined based on a typical state where the ship is anchored and only maintains basic loads such as lighting and ventilation.

[0034] Third, continuously monitor the ship's position coordinates and calculate the position change between two consecutive sampling points. Calculate the distance between two adjacent coordinates using the spherical distance formula, taking into account the Earth's curvature. When the maximum position change in ten consecutive samplings is less than one hundred meters, the ship is considered to be stationary.

[0035] When all three conditions above are met simultaneously, and the stationary state lasts for more than 30 minutes, the vessel is determined to be in anchorage condition. The stationary duration is used to rule out situations of temporary anchoring or waiting for pilotage, confirming that the vessel is stably anchored.

[0036] After determining that the ship is in a condition where power outage is permissible, a test request is issued to the operators, displaying the current operating status and a list of circuits eligible for testing. Non-critical circuits that can be temporarily disconnected are selected from the ship's electrical distribution system. Disconnectable circuits include: some cabin lighting circuits, some ventilation circuits, some auxiliary equipment power supply circuits such as deck machinery power supply, and some domestic power supply circuits. Critical circuits are not included in the test; critical circuits include: steering gear power, main engine control power, navigation equipment power, fire protection system power, emergency lighting power, and communication equipment power.

[0037] Before the test begins, record the reference value of the total system-to-ground capacitance displayed by the current IT system insulation monitoring device. This value is calculated by the insulation monitoring device by measuring the system's admittance to ground by injecting a test signal of a specific frequency into the system, and then calculating the total value of the capacitance to ground.

[0038] Disconnect the circuit breakers of each disconnectable branch in sequence, and the insulation monitoring device remeasures and displays the capacitance to ground value. Record the capacitance to ground value before disconnection and the capacitance to ground value after disconnection; the difference between the two is the change in capacitance to ground for that branch. Theoretically, this change is equal to the capacitance to ground value of the cable in the disconnected branch.

[0039] Obtain the cable length information of each branch, divide the change in capacitance to ground of each branch by the corresponding cable length to obtain the capacitance to ground value per unit length, which is the normalized ratio.

[0040] Data on the unit-length capacitance-to-ground ratios of all participating test branches were collected to form a dataset. Outlier removal was first performed to eliminate invalid data caused by measurement errors or extreme anomalies. The initial mean and initial standard deviation of all ratios were calculated. The deviation of each ratio from the initial mean was divided by the initial standard deviation to obtain the standardized deviation. Data points with an absolute value of standardized deviation greater than five were considered extreme outliers and removed. The outlier removal factor was set to five standard deviations.

[0041] After removing outliers, the mean and standard deviation are recalculated based on the remaining data.

[0042] Check if the capacitance-to-ground ratio per unit length of each branch is within the normal range. If the ratio of a branch is less than the lower limit of the normal range, it indicates that the capacitance-to-ground of the cable in that branch is abnormally reduced, possibly due to a cable break or measurement error. If the ratio of a branch exceeds the upper limit of the normal range, it indicates that the capacitance-to-ground of the cable in that branch is abnormally increased, and it is identified as a faulty branch. Physical causes of increased capacitance-to-ground include: moisture absorption of the insulation layer leading to an increase in dielectric constant, local compression of the insulation layer leading to a decrease in thickness, and the formation of dendritic discharge channels within the insulation layer.

[0043] The normal range definition factor is set at 3 times the standard deviation, which is based on the statistical characteristics of the normal distribution.

[0044] After locating the faulty branch, power supply to all branches is restored in sequence, i.e., the previously disconnected circuit breakers are closed, and the system returns to normal operation.

[0045] After power is restored to all branches, data on leakage current to ground and phase voltage are specifically collected for the identified faulty branches. The data acquisition devices remain zero-sequence current transformers and voltage transformers, but the focus is now on the signals from the faulty branch.

[0046] A Fast Fourier Transform (FFT) is performed on the acquired leakage current time-domain waveform to convert the time-domain signal into a frequency-domain signal. The amplitude and phase angle of the fundamental component of the phase voltage are extracted simultaneously. The phase of the fundamental phase voltage is typically selected with its zero-crossing point as the reference phase.

[0047] The phase difference is obtained by calculating the difference between the fundamental phase angle of the leakage current and the fundamental phase angle of the phase voltage. A positive phase difference indicates that the current leads the voltage, while a negative phase difference indicates that the current lags the voltage. In IT systems, under normal circumstances, the leakage current is mainly capacitive leakage, with the current leading the voltage by 90 degrees and the phase difference being positive 90 degrees.

[0048] Phase difference reflects the nature of leakage current. In pure capacitive leakage, the leakage current is entirely generated by the capacitance to ground, and the relationship between current and voltage is: current equals angular frequency multiplied by capacitance multiplied by voltage, with the current leading the voltage by 90 degrees. In pure resistive leakage, the leakage current is generated by the insulation resistance, and the current is proportional to the voltage and in phase, with a phase difference of zero degrees. In mixed leakage, the leakage current contains both capacitive and resistive components, with a phase difference between zero and 90 degrees. The closer the phase difference is to zero degrees, the greater the proportion of the resistive component, and the more severe the insulation degradation. Phase difference reflects the capacitive-resistive ratio; a phase angle shift from 90 degrees to 0 degrees indicates an increase in resistive leakage.

[0049] Spectral analysis was performed on the time-domain waveform of the leakage current to extract the amplitude of each harmonic component, in addition to the fundamental component. The harmonic orders included the second, third, and up to the fifteenth harmonic. The amplitudes of harmonics above the fifteenth order were usually very small, contributing little to the harmonic distortion rate, and were susceptible to high-frequency noise interference.

[0050] Calculate the sum of squares of the amplitudes of the second to fifteenth harmonics, and take the square root of this sum to obtain the total harmonic amplitude. Divide the total harmonic amplitude by the fundamental frequency amplitude to obtain the total harmonic distortion (THD), which reflects the degree to which the current waveform deviates from a pure sine wave. The THD reflects the nonlinear breakdown characteristics of the leakage path.

[0051] The fault type is determined based on the combined characteristics of phase difference and harmonic distortion rate. The determination rules are as follows: Mild insulation degradation: The phase difference is between 90 and 70 degrees capacitive lead, and the harmonic distortion rate is below 5%. This state indicates that the insulation resistance has begun to decrease, and a small amount of resistive leakage current has appeared, but the resistive component is still relatively small compared to the capacitive component, and the phase difference has only shifted from 90 degrees to 70 degrees. A harmonic distortion rate below 5% indicates that the leakage path still exhibits linear component characteristics, and nonlinear breakdown has not yet occurred. The insulation resistance is typically in the range of hundreds of megohms to gigaohms, indicating the early stages of degradation.

[0052] The determination of the first angle threshold: 70 degrees corresponds to a ratio of capacitive impedance to resistive impedance of approximately 2.7 to 1, that is, the capacitive component is 2.7 times the resistive component. This angle range is based on the typical characteristic of leakage current changing from purely capacitive to mixed characteristics when the insulation resistance decreases from extremely high values ​​(gigaohms) to relatively high values ​​(hundreds of megaohms).

[0053] Determination of the first harmonic distortion rate threshold: 5% corresponds to a current waveform close to a pure sine wave. This threshold is determined based on measured data of normal capacitive leakage current. Under normal circumstances, the harmonic distortion rate of capacitive leakage current is typically in the range of 1% to 3%.

[0054] Moderate insulation degradation: The phase difference is between 70 and 45 degrees capacitive lead, and the harmonic distortion rate is less than 10%. This state indicates that resistive leakage current accounts for a large proportion, and the phase difference has further shifted from 70 degrees to 45 degrees. A 45-degree phase difference means that the resistive component is equivalent to the capacitive component, that is, the capacitive reactance and impedance amplitudes are equal. A harmonic distortion rate of less than 10% indicates that although the insulation resistance of the leakage path has decreased significantly, it still maintains linear characteristics and has not broken down.

[0055] The second angle threshold is determined as follows: 45 degrees corresponds to the critical state where the capacitive and resistive components are equal. This angle indicates that the insulation has entered a state of moderate degradation and requires close monitoring. The angle threshold can be adjusted within the range of 30 to 60 degrees depending on the cable type and maintenance strategy.

[0056] Determination of the second harmonic distortion rate threshold: 10% corresponds to the presence of perceptible harmonic components in the current waveform, but not yet reaching the level of severe distortion.

[0057] Insulation breakdown: The phase difference is between 45 degrees leading capacitively and in phase, and the harmonic distortion rate is higher than 15%. A phase difference close to zero degrees indicates that the resistive component dominates the leakage current, and the capacitive component is negligible. A harmonic distortion rate higher than 15% indicates that the leakage path has partially broken down, forming a nonlinear leakage path; the high distortion rate reflects the nonlinear arc discharge generated at the breakdown point. Breakdown may manifest as the formation of carbonized channels within the insulation layer, air gap discharge, or metallic short circuit. The voltage-current relationship at the breakdown point exhibits nonlinear characteristics, generating abundant harmonic components under sinusoidal voltage excitation. This state indicates a serious fault and requires immediate attention.

[0058] Determining the high threshold for harmonic distortion rate: 15% corresponds to a current waveform that has undergone significant distortion and no longer closely approximates a sine wave. This threshold is based on measured data of local breakdown and arc discharge characteristics. Setting the threshold to 15% allows for fault identification in the early stages of breakdown.

[0059] Intermittent faults: Abrupt changes are detected in the time-domain waveform of the leakage current. A sliding window method is used, with a window length of ten sampling points. The current difference between adjacent sampling points within the window is calculated, i.e., the current value of the later sampling point minus the current value of the earlier sampling point. Simultaneously, the standard deviation of the current within the window is calculated as a measure of normal fluctuation amplitude.

[0060] When the absolute value of a current difference exceeds five times the standard deviation within the window, that point is considered a sudden change. Five times the standard deviation corresponds to a significantly abnormal change, effectively distinguishing between true sudden changes and normal noise fluctuations. The number of sudden changes detected within one minute is counted; when the frequency of sudden changes exceeds ten times, it is classified as an intermittent fault.

[0061] Determining the abrupt change frequency threshold: This frequency is much higher than the frequency of occasional electromagnetic interference or operational disturbances, and can effectively distinguish between intermittent faults and occasional noise interference. Intermittent faults are usually caused by poor conductor contact or insulation cracks due to vibration, and their abrupt change frequency is related to the vibration frequency.

[0062] When the faulty branch enters the power outage maintenance phase, precise positioning is implemented; first, the total cable length of the faulty branch is obtained, and then the minimum frequency of the required high-frequency signal is calculated according to the positioning accuracy requirements.

[0063] The required positioning accuracy is determined based on the typical spacing of critical locations along the ship's cable route. Critical locations include branch junction boxes, hull-penetrating seals, and cable support fixing points, which are high-risk areas for failure and are easily accessible for maintenance. The spacing between adjacent critical locations is typically between 10 and 50 meters.

[0064] The propagation speed of high-frequency signals in a cable depends on the cable's distributed inductance and distributed capacitance. The propagation speed is equal to the reciprocal of the square root of the product of the distributed inductance and distributed capacitance.

[0065] According to the principle of time-domain reflectometry, the positioning accuracy is approximately equal to half the signal wavelength; wavelength equals propagation speed divided by frequency. Therefore, the minimum frequency equals propagation speed divided by twice the positioning accuracy.

[0066] In addition to the minimum frequency required to achieve positioning accuracy, a frequency must be selected that avoids the main interference bands of the ship's electrical system as the actual carrier frequency. The main interference bands of the ship's electrical system include: First, the power frequency of 50 Hz and its lower harmonics, including 100 Hz, 150 Hz, 200 Hz, 250 Hz, 300 Hz, etc.

[0067] Second, the switching frequency and sidebands of the frequency converter. The pulse width modulation signal generated during the switching process of the frequency converter contains rich spectral components, which create sideband interference near the carrier frequency.

[0068] Third, the high-frequency bands of radar and communication equipment. Since the frequency of cable detection signals is much lower than that of radar and communication frequencies, and radar signals propagate through space rather than being conducted through cables, the interference of this frequency band on cable detection is negligible.

[0069] Taking into account both positioning accuracy and interference avoidance, a frequency in the range of 50 kHz to 200 kHz was selected as the carrier frequency. This frequency band is much higher than the power frequency and low-order harmonics, and also higher than the inverter carrier frequency, which can effectively avoid the main sources of interference.

[0070] Select accessible key locations along the cable path as measurement points. The selection of measurement locations follows these principles: First, the cable start-up point, i.e., the power distribution switch, serves as the high-frequency signal injection point and the first measurement point. The start-up point is fixed and easily accessible, allowing for the installation of a signal source and measuring instruments.

[0071] Second, the location of each branch junction box; branch junction boxes are usually located at the nodes where the power distribution trunk line branches to each electrical equipment. These locations are equipped with terminals, which makes it easy to open the box cover for measurement without damaging the cable insulation.

[0072] Third, the point where the cable passes through the watertight bulkhead; this is where the cable passes through the sealing device, where the cable is subjected to mechanical compression and the insulation layer is compressed, making it a high-risk point for failure. The point of entry into the bulkhead is usually marked on both sides of the bulkhead for easy location.

[0073] Fourth, cable bracket fixing points; the location where cables are fixed to the ship's structure by clamps or brackets. Cables at this location are greatly affected by vibration, and conductor fatigue and insulation wear are common failure modes.

[0074] Fifth, the cable end, i.e., the terminal of the electrical equipment, serves as the last measurement point. The end position marks the termination of the cable, completing the entire measurement process.

[0075] The interval between measurement points is determined based on the required positioning accuracy. If the required positioning accuracy is 20 meters, the interval between adjacent measurement points should not exceed 20 meters to ensure that the fault point can be located within a certain segment between two adjacent measurement points. Under the constraints of the total cable length and the number of accessible locations, high-fault locations (such as passageways and fixed points) should be prioritized as measurement points, and temporary measurement points should be added as necessary to meet the interval requirements.

[0076] A high-frequency carrier signal is injected at the cable end through a coupling capacitor. The function of the coupling capacitor is to couple the high-frequency signal source to the cable while blocking DC and power frequency components, thus preventing the signal source from affecting the DC insulation test of the cable. The capacitance of the coupling capacitor should be selected so that its capacitive reactance at the carrier frequency is much smaller than the cable impedance to ensure that the signal is effectively injected into the cable.

[0077] The high-frequency signal source outputs a sinusoidal signal at a pre-selected carrier frequency. The amplitude and power level are sufficient to allow the signal to propagate along the cable for hundreds to thousands of meters while maintaining a measurable amplitude. At the same time, the power is low enough not to cause thermal damage or insulation breakdown to the cable after power is cut off.

[0078] The amplitude of the high-frequency carrier current was measured at various measurement locations using a portable high-frequency current clamp meter. The high-frequency current flowing through the cable generates an alternating magnetic flux in the magnetic core, which induces a voltage in the induction coil. The voltage amplitude is proportional to the current amplitude. The instrument amplifies and detects the induced voltage, displaying the effective value of the high-frequency current.

[0079] Record the amplitude of the injected current at the beginning of the cable, and denote it as the first measuring point current. Measure the amplitude of the high-frequency current at each measuring point set sequentially along the cable direction, and denote it as the second measuring point current, third measuring point current, and so on.

[0080] Calculate the current attenuation rate between adjacent measuring points. The attenuation rate from the first measuring point to the second measuring point is equal to the difference between the current at the first measuring point and the current at the second measuring point, divided by the current at the first measuring point. Continue this process to calculate the attenuation rate between all adjacent measuring points.

[0081] Under normal circumstances, as high-frequency current propagates along a cable, the ohmic loss due to distributed resistance and the dielectric loss of the insulation medium gradually decrease. The attenuation follows an exponential decay characteristic, with the current amplitude decreasing exponentially with distance.

[0082] Establishing a baseline for normal attenuation rate: During ship construction or overhaul, high-frequency current attenuation tests are performed on normal cable sections. A known normal cable is selected, and a signal is injected at the same carrier frequency to measure the attenuation rate of different length sections.

[0083] Compare the actual attenuation rate of each measurement segment with the normal attenuation rate benchmark to identify segments with abnormally high attenuation rates. If the attenuation rate of a segment exceeds twice the normal attenuation rate, a preliminary assessment indicates a possible fault in that segment. If the attenuation rate of a segment exceeds three times the normal attenuation rate, a fault in that segment is highly suspected.

[0084] Determining the threshold for abnormal attenuation rate: The additional attenuation caused by a fault point is usually significantly greater than the normal attenuation. Actual measurement data shows that the attenuation rate of a faulty section is typically 3 to 10 times that of a normal section. Setting the threshold to 2 to 3 times can effectively identify faults while avoiding misjudging minor anomalies as serious faults.

[0085] By setting measuring points along the cable segments, calculating the current attenuation rate of each segment and comparing it with a benchmark value, and progressively increasing the density of measuring points in abnormal sections, the fault range is narrowed down. Finally, combined with the cable laying path diagram, the specific physical location is determined, achieving precise location of the cable fault. Maintenance personnel arrive at the site based on this location information and focus on inspecting the cable at the cableway penetration point and within a five-meter radius before and after it.

[0086] After determining that an insulation degradation fault exists, the amplitude change of the neutral-to-ground voltage in the IT system is continuously monitored. The neutral-to-ground voltage is a parameter unique to IT systems, reflecting the degree of imbalance in the three-phase impedance to ground. The neutral point of an IT system is usually ungrounded or grounded through a high-resistance connection, and a voltage exists between the neutral point and ground.

[0087] The amplitude of the neutral point-to-ground voltage is measured by an insulation monitoring device. This device connects a high-resistance voltmeter or voltage transformer between the neutral point and ground to measure the voltage of the neutral point relative to the ship's ground. The change in the neutral point-to-ground voltage over time is recorded to form time-series data.

[0088] Analyze the trend of neutral-to-ground voltage variation. Calculate the linear trend of the voltage time series, i.e., fit a straight line using the least squares method. The slope of the line reflects the voltage variation trend. A positive slope indicates a monotonically increasing voltage, a negative slope indicates a monotonically decreasing voltage, and a slope close to zero indicates that the voltage is basically stable.

[0089] Simultaneously, the variation trend of the active component of the leakage current was analyzed. Using the same method, a linear fit was performed on the time series of the active component to obtain the slope of its variation trend.

[0090] The correlation coefficient between the neutral point-to-ground voltage variation trend and the active component variation trend of the leakage current was calculated. The Pearson correlation coefficient method was used, with the correlation coefficient ranging from -1 to +1. A correlation coefficient close to +1 indicates a high positive correlation between the two variables, i.e., synchronous growth; a correlation coefficient close to -1 indicates a high negative correlation, i.e., inverse changes; and a correlation coefficient close to zero indicates no linear correlation.

[0091] When the neutral-to-ground voltage shows a monotonically increasing trend (the fitting slope is significantly greater than zero) and is synchronized with the growth trend of the active component of the leakage current, that is, when the Pearson correlation coefficient between the two is greater than 0.8, it is used as an auxiliary basis for confirming insulation degradation faults.

[0092] Determining the correlation coefficient threshold: A Pearson correlation coefficient between 0.7 and 0.9 is generally considered a strong correlation. Considering the presence of noise and other interference factors in actual measurements, the threshold is set at 0.8.

[0093] This method leverages the consistency between changes in global parameters (neutral point voltage) and local parameters (leakage current) of an IT system. Insulation degradation leads to increased leakage to ground in a particular branch, causing a decrease in the branch's impedance to ground. This affects the balance of the three-phase impedance to ground, resulting in a change in the neutral point voltage. The synchronized trends of both parameters confirm the accuracy of the fault diagnosis, avoiding potential misjudgments that might occur with relying on a single parameter.

[0094] Temperature sensors are installed at key compartment locations along the cable route to collect ambient temperature data. Key locations include representative points in different temperature environments, such as the engine room, cargo hold, and superstructure. The engine room temperature is typically higher due to the operation of the main and auxiliary engines. The cargo hold temperature is close to the ambient temperature and varies with the season and route. The superstructure has air conditioning, resulting in a relatively stable temperature.

[0095] Temperature sensors are installed in at least three different compartment locations along each monitoring cable. The sensors are mounted near the cable but not in direct contact with it to avoid being affected by the cable's own heat. The sensors are connected to the data acquisition system via signal lines to transmit temperature data in real time.

[0096] A statistical model of the relationship between the active component of leakage current and ambient temperature was established. Leakage current data were collected under different temperature conditions during normal ship operation. Ambient temperature varies significantly when the ship operates on different routes and in different seasons; these naturally changing temperature conditions were utilized for data collection. Ambient temperature and the active component of leakage current were recorded simultaneously to form a paired dataset.

[0097] Regression analysis is performed on the paired data to establish the relationship curve between the active power component and temperature. If the active power component and temperature exhibit a linear relationship, linear regression is used to fit the linear equation and obtain the slope and intercept parameters. If a nonlinear relationship is observed, polynomial regression is used; typically, a quadratic polynomial is sufficient to meet the accuracy requirements. After the regression model is established, the model parameters are stored in the detection system.

[0098] During actual testing, based on the measured ambient temperature, the expected leakage current value based on the temperature effect is calculated using an established regression model. The active component of the actual measured leakage current is compared with the expected value, and the ratio is calculated. If the actual value is close to the expected value, and the ratio is within the range of 0.8 to 1.2, it indicates that the change in leakage current is mainly caused by temperature fluctuations, which is a normal phenomenon. If the actual value significantly exceeds the expected value, and the ratio exceeds 1.5 to 2 times, the excess portion is determined to be abnormal leakage caused by insulation degradation, rather than normal fluctuations caused by changes in ambient temperature.

[0099] It effectively separates the effects of ambient temperature changes and insulation degradation on leakage current, improving the accuracy of fault diagnosis and reducing false alarms caused by seasonal temperature changes or changes in the flight path environment.

[0100] Determining the number of temperature measurement points: Each monitoring cable should have at least three measurement points to ensure coverage of the different temperature zones the cable traverses. These three points correspond to the cable's starting point, midpoint, and ending point, or the three main compartments. If the cable path is complex and crosses many compartments, the number of measurement points can be increased to 5 to 10. More measurement points result in higher temperature compensation accuracy, but also increase system complexity and cost. A balance between accuracy and cost should be struck based on actual application requirements.

[0101] A triaxial vibration acceleration sensor is installed at the ship's main engine base to collect vibration signals. The triaxial sensor can simultaneously measure vibration acceleration in the horizontal, lateral, and vertical directions. The main engine base is the primary source of ship vibration, and the vibration signals from this location are representative.

[0102] The sensor is fixed to a rigid part of the main unit base by bolts or a magnetic base, ensuring that the sensor vibrates synchronously with the base and accurately measures the vibration signal. The sensor outputs acceleration signals through three channels, which are amplified and filtered by a signal conditioning circuit before being converted into digital signals by an analog-to-digital converter. The sampling frequency is set to 1,000 times per second, which is sufficient to capture the operating frequency of the main unit and its harmonics.

[0103] Spectral analysis of the vibration signal is performed using the Fast Fourier Transform (FFT) method to convert the time-domain vibration signal into a frequency-domain signal. The frequency component with the largest amplitude in the spectrum is extracted; this frequency is the dominant frequency of the vibration. The dominant frequency is usually related to the rotational speed of the host machine.

[0104] At the same time, the vibration amplitude corresponding to the main frequency is extracted, and the amplitude reflects the intensity of the vibration.

[0105] Synchronously acquire the time-domain waveforms of leakage current or capacitance to ground. The sampling frequency is the same as that of the vibration signal, maintained at 1,000 times per second, to ensure time alignment. Perform spectral analysis on the time-domain waveforms of leakage current or capacitance to ground to obtain their frequency domain characteristics.

[0106] Compare the spectrum of the leakage current or capacitance to ground with the spectrum of the vibration signal. Check if there is a frequency component in the spectrum of the leakage current or capacitance that matches the dominant vibration frequency. The criterion for frequency matching is a frequency deviation of less than 1 Hz.

[0107] Determining the frequency deviation threshold: Considering that the host's rotational speed fluctuates slightly during stable operation, typically within ±2% of the rated speed, corresponding to a fluctuation of tens of revolutions per minute. Simultaneously, the frequency resolution of the spectrum analysis is considered; the frequency resolution equals the sampling frequency divided by the number of sampling points.

[0108] Further calculations are performed to determine the frequency domain correlation coefficient between the leakage current or capacitance spectrum and the vibration spectrum near the dominant frequency. The frequency domain correlation coefficient is obtained by calculating the cross-correlation function of the two spectra within a certain bandwidth near the dominant frequency (e.g., the dominant frequency plus or minus 5 Hz). A correlation coefficient greater than 0.7 indicates a significant correlation between the two signals at that frequency component.

[0109] Determining the threshold for the frequency domain correlation coefficient: This threshold takes into account that in addition to the vibration-related component, there are other noise and interference components in the actual signal, resulting in a less than perfect correlation. Setting the threshold to 0.7 can both identify real vibration-related faults and avoid misjudging weakly correlated incidental phenomena.

[0110] Temperature sensors are deployed in critical areas such as the engine compartment to establish a statistical model of leakage current versus ambient temperature, subtracting the natural decrease in insulation resistance due to temperature rise. Triaxial vibration signals from the main engine base are collected, and the frequency domain correlation between electrical fluctuations and mechanical vibrations is analyzed using coherence functions. When the dominant frequencies of both are consistent and the coherence is greater than 0.7, it is determined to be an intermittent fault caused by poor conductor contact due to vibration, rather than material degradation itself.

[0111] When the frequency of leakage current or ground capacitance fluctuation is detected to be consistent with the dominant vibration frequency (frequency deviation less than one Hz) and the frequency domain correlation coefficient exceeds 0.7, it is determined to be a vibration-induced intermittent fault. The specific procedure is as follows: Figure 3 This fault is typically caused by conductor fatigue fracture or loose joints. Under vibration, the conductor contact at the fracture point is intermittent, or the contact resistance of the loose joint changes periodically, causing electrical parameters to fluctuate with the vibration frequency. This method, through frequency domain correlation analysis, links the characteristics of electrical faults with the characteristics of mechanical vibration, accurately identifying intermittent faults caused by vibration and providing a strong basis for fault cause diagnosis.

[0112] Example 2: Based on Example 1, this embodiment focuses on tracking the insulation degradation process based on load switching transient analysis, fault confirmation based on inter-branch waveform comparison, and subdivision of intermittent fault types.

[0113] The specific scheme for tracking the insulation degradation process based on load switching transient analysis is as follows: During ship operation, the switching action of high-power loads is continuously monitored. When a load switching signal is detected, the transient waveform of the leakage current to ground before and after the load switching is acquired at high speed. Under the condition that the system resistance to ground changes relatively slowly, the time constant of the discharge process of the capacitance to ground is extracted. By statistically analyzing the time constants extracted during multiple load switching processes, the trend of the time constant is calculated. When the time constant shows a monotonically increasing trend and the growth rate exceeds a predetermined threshold, it is determined that the cable capacitance to ground continues to increase, which serves as an auxiliary criterion for accelerating the insulation degradation process.

[0114] Specifically, during normal ship operation, the detection system continuously monitors the load switching signals output by the ship's automation system. The operation of high-power loads on the ship, such as the main pump unit, air compressor, and propulsion motor, is intermittent. This is achieved by receiving the tripping signals from the auxiliary contacts of the load circuit breaker.

[0115] When a load switching signal is detected, the data acquisition module switches to high-speed acquisition mode to acquire the ground leakage current waveform.

[0116] At the instant the load is disconnected, the charge stored in the cable's distributed capacitance to ground discharges through the system's equivalent resistance to ground. In IT systems, the system's equivalent resistance to ground is mainly composed of the injection resistance of the insulation monitoring device and the cable insulation resistance connected in parallel. The capacitance to ground discharges through this resistance, and the discharge current exhibits an exponential decay characteristic. Utilizing the exponential characteristic of capacitance discharging through the system's equivalent resistance, the discharge time constant is extracted through curve fitting.

[0117] The data processing unit performs exponential function fitting on the acquired discharge current waveform. The fitting employs the least squares method to find the optimal time constant parameter that minimizes the mean square error between the fitted curve and the measured waveform. The fitted time constant reflects the product of the system's equivalent resistance to ground and the cable's capacitance to ground.

[0118] In the early and middle stages of insulation degradation, the cable insulation resistance is still greater than 10 megohms, far exceeding the injection resistance of the insulation monitoring device. At this time, the system's equivalent resistance to ground is mainly determined by the injection resistance, and its influence from cable insulation degradation is less than 5%. Under these conditions, the change in time constant is mainly attributed to the change in ground capacitance. When insulation degradation progresses to the later stage, the insulation resistance drops to the megohm level, becoming comparable to the injection resistance. At this point, the system's ground resistance begins to decrease significantly. The time constant is then affected by both the system's ground resistance and ground capacitance. However, by this stage, insulation degradation is already quite severe, and the early warning target of this method has been achieved.

[0119] A time constant database was established using a relational data structure. Each record contains five fields: branch number, load switching timestamp, time constant value, estimated system-to-ground resistance, and data validity flag. The database is indexed by branch number to facilitate quick retrieval of historical data for specific branches.

[0120] Trend analysis employs a moving window linear regression method. During moving window analysis, based on the branch number and the current time, the most recent 20 to 50 valid records for that branch are extracted from the database, sorted in ascending order by timestamp, to form a time series array.

[0121] The predetermined threshold for the rate of change of the time constant is determined through benchmark testing. During long-term monitoring of normal cables at the time of ship delivery or after cable replacement, time constant data are recorded for a period of three to six months. The maximum rate of change of the time constant due to natural fluctuations during this period is calculated. This value reflects normal fluctuations caused by environmental factors such as temperature and humidity, as well as measurement errors. This maximum rate of change of normal fluctuation is recorded as the benchmark rate of change.

[0122] When the rate of change of the real-time calculated time constant exceeds the threshold, and the moving window shows that the rate of change exceeds the threshold in three consecutive updates, it is determined that the time constant exhibits a monotonically increasing trend. Monotonically increasing time constant indicates a continuous increase in capacitance to ground, corresponding to continuous insulation degradation. The magnitude of the rate of change reflects the rate of degradation; the larger the rate of change, the faster the degradation process. This judgment result is used as an auxiliary criterion for accelerating the insulation degradation process, combined with the primary criterion based on the correlation between active power components and load current, to achieve an improvement in functionality from fault identification to degradation process prediction. Simultaneously, waveforms of the faulty branch and normal branches of the same model are synchronously acquired and compared laterally using Pearson correlation. If the similarity is less than 0.75, common system interference factors are excluded, confirming the effectiveness of fault location.

[0123] The specific scheme for fault confirmation based on inter-branch waveform comparison is as follows: After power is restored to each branch, the zero-sequence current waveforms of the faulty branch and at least two normal branches are collected synchronously. Time-domain features are extracted from the zero-sequence current waveforms of each branch, and the similarity between the waveform of the faulty branch and the waveforms of each normal branch is calculated. When the similarity between the waveform of the faulty branch and the waveforms of all normal branches is lower than a predetermined similarity threshold, the validity of the faulty branch judgment is confirmed. When the similarity between the waveform of the faulty branch and the waveform of at least one normal branch is higher than the predetermined similarity threshold, it is marked as a suspected misjudgment, triggering a manual review process.

[0124] Specifically, after locating the faulty branch, a cross-branch comparison is performed to improve the reliability of the judgment. This confirmation process is executed immediately after power is restored to each branch.

[0125] The data acquisition system is equipped with multi-channel synchronous acquisition functionality. At the main distribution board, a zero-sequence current transformer is installed for each distribution branch. The output signals of each transformer are synchronously acquired via a multi-channel data acquisition card, with a unified sampling clock to ensure time alignment of signals from each channel.

[0126] The synchronous data acquisition duration is set to 1 to 3 minutes, and the sampling frequency is maintained at 1 kilohertz to 10 kilohertz per second. The data acquisition targets include the located faulty branch and at least two normal branches. The selection criteria for normal branches are: the cable model is the same as or similar to that of the faulty branch, the rated current level is the same, and the load type is similar to ensure the fairness of the comparison.

[0127] Time-domain features are extracted from the zero-sequence current waveforms of each branch. These time-domain features include the waveform's mean, standard deviation, peak value, peak-to-peak value, zero-crossing rate, and autocorrelation function characteristics, which characterize the statistical and dynamic properties of the waveforms from different perspectives.

[0128] Calculate the similarity between the waveform of the faulty branch and the waveforms of each normal branch. The Pearson correlation coefficient method is preferred for similarity calculation due to its simplicity and insensitivity to waveform amplitude scaling. When the lengths of the two waveforms being compared are unequal, or when there is a significant time shift, the dynamic time-warped distance method is used instead. After calculating the dynamic time-warped distance, the distance value is divided by the sum of the amplitude ranges of the two waveforms to obtain the normalized distance. The similarity is defined as one minus the normalized distance, ensuring that the similarity value range is consistent with the Pearson correlation coefficient, ranging from -1 to +1.

[0129] The predetermined similarity threshold is set to 0.75. This threshold is based on the following considerations: Although the zero-sequence current waveforms between normal branches are generated by different cables, they are mainly capacitive leakage, and their waveform characteristics are similar, with a similarity to each other typically greater than 0.8. Faulty branches, due to the presence of abnormal resistive leakage or breakdown components, exhibit significantly altered waveform characteristics, and their similarity to normal branches is typically lower than 0.6.

[0130] The judgment logic is as follows: When the similarity between the waveform of the faulty branch and the waveforms of all normal branches is less than 0.75, it indicates that the waveform characteristics of the faulty branch are significantly different from all normal branches, confirming the validity of the faulty branch judgment. When the similarity between the waveform of the faulty branch and the waveform of at least one normal branch is greater than 0.75, the system marks it as a suspected misjudgment, triggering a manual review process where professionals make a comprehensive judgment based on other information. This horizontal comparison method uses normal branches as a dynamic reference benchmark, which can adapt to changes in current operating conditions and environmental conditions, improving the accuracy and robustness of the judgment.

[0131] The specific details of the intermittent fault type subdivision are as follows: After identifying the intermittent fault, statistical analysis is performed on the amplitude and duration of the leakage current sudden pulse, and histograms of amplitude distribution and duration distribution of the sudden pulse are established. The triggering mechanism of the intermittent fault is determined based on the dispersion of amplitude distribution and the concentration of duration distribution. When the dispersion of amplitude distribution of sudden pulse is high and the concentration of duration distribution is high, it is identified as a poor contact type intermittent fault; when the dispersion of amplitude distribution of sudden pulse is low and the dispersion of duration distribution is high, it is identified as an insulation breakdown type intermittent fault.

[0132] Specifically, after identifying intermittent faults, the triggering mechanism of intermittent faults is further refined to distinguish between two types: poor contact and insulation breakdown.

[0133] Statistical analysis was performed on the detected leakage current abrupt changes. Each abrupt change pulse contained two key parameters: amplitude and duration. The amplitude was defined as the absolute value of the difference between the current value at the abrupt change point and the reference current value before the abrupt change. The reference value was determined using a moving average method, calculating the average current value of the 10 sampling points before the abrupt change. The duration was defined as the time span from the start of the abrupt change to the current recovering to the reference value plus or minus three standard deviations.

[0134] If fewer than 20 mutation pulses are detected within the observation period, the statistical sample is insufficient to establish a reliable distribution characteristic. In this case, the system extends the observation period to 10 minutes or longer until at least 20 mutation pulses are accumulated. If the minimum sample size is not reached even after extending the observation period to 30 minutes, the system determines it to be a low-frequency intermittent fault, and does not further classify the type; it is simply marked as an intermittent fault for further observation.

[0135] After accumulating more than 20 abrupt pulses, frequency distribution histograms are constructed for the amplitude and duration of all pulses. The horizontal axis of the amplitude distribution histogram represents the amplitude range, dividing the range from the maximum amplitude to the minimum amplitude into 10 to 20 equal intervals. The vertical axis represents the number of pulses falling within each interval.

[0136] The dispersion of the amplitude distribution is calculated using the coefficient of variation, defined as the standard deviation divided by the mean. The central tendency of the duration distribution is calculated using the kurtosis index, which measures the sharpness of the distribution.

[0137] The discrimination rule is based on a combination of amplitude dispersion and duration concentration. For cross-linked polyethylene insulated cables, when the coefficient of variation of the amplitude distribution is greater than 0.7 and the kurtosis of the duration distribution is greater than four, it is classified as a poor contact type intermittent fault. When the coefficient of variation of the amplitude distribution is less than 0.7 and the kurtosis of the duration distribution is less than four, it is classified as an insulation breakdown type intermittent fault. The same discrimination threshold applies to rubber insulated cables.

[0138] For intermittent faults, the amplitude and duration of the sudden pulses are statistically analyzed. A distribution characteristic histogram is established: when the pulse amplitude is highly discrete and the duration is highly concentrated within a specific vibration cycle, it is identified as a contact failure such as a loose joint; when the pulse amplitude is relatively concentrated near the breakdown voltage and the duration is randomly distributed, it is identified as an air gap breakdown failure within the material.

[0139] The physical mechanism of poor contact faults is as follows: under vibration, the contact resistance at the conductor connection changes randomly, resulting in highly discrete amplitudes of sudden pulses; while the contact duration is mainly controlled by the vibration period, and the vibration frequency is relatively stable, resulting in a highly concentrated duration. The physical mechanism of insulation breakdown faults is as follows: the breakdown voltage is relatively stable, and the sudden amplitude is concentrated; while the duration of the breakdown channel is affected by various random factors such as temperature and humidity, exhibiting high dispersion.

[0140] Detailed fault type identification provides targeted guidance for maintenance decisions. For contact-related faults, the focus of maintenance is on tightening loose connections or replacing oxidized terminals. For insulation breakdown faults, the damaged cable section needs to be replaced. Accurate identification can avoid incorrect maintenance and improve efficiency.

[0141] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An automatic detection method for online fault diagnosis of marine cables, characterized in that, include: The system collects the ground leakage current and load current during operation, extracts the active component of the leakage current, analyzes the correlation between the active component and the load current, and determines that there is an insulation degradation fault when the active component is nonlinearly positively correlated with the load current. When the ship is in a condition where short-term power outages of some non-critical circuits are permitted, the capacitance to ground of each power-disconnectable branch is obtained before and after sequential disconnection. The change in capacitance to ground of each branch is calculated, and the change in capacitance is normalized with the corresponding branch cable length. Branches whose normalized values ​​exceed the normal range are identified as faulty branches. After power is restored to each branch, the leakage current to ground and phase voltage are collected for the faulty branch. The phase difference between the fundamental component of the leakage current and the voltage and the harmonic distortion rate are extracted. Based on the combined characteristics of the phase difference and the harmonic distortion rate, the fault type is determined to be insulation deterioration, insulation breakdown and / or intermittent fault. When the faulty branch is de-energized for maintenance, a high-frequency carrier signal is injected into the cable, and the high-frequency current amplitude at multiple accessible locations along the cable is measured. The fault point is determined based on the location segment where the high-frequency current attenuation rate between adjacent measuring points increases abnormally.

2. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that: Extracting the active component of leakage current includes: performing spectral decomposition on the collected leakage current to obtain the fundamental component, extracting the component of the fundamental component that is in phase with the phase voltage as the active component, and extracting the component of the fundamental component that is orthogonal to the phase voltage as the reactive component. The analysis of the correlation between active power components and load current includes: synchronously recording time series data of active power components and load current within a continuous time window, calculating the rate of change of active power components with respect to load current, and determining that when the rate of change is greater than a first preset threshold and maintains a monotonically increasing trend within multiple continuous time windows, it is a nonlinear positive correlation.

3. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that: The ship is determined to be in a condition where a short-term power outage of some non-critical circuits is permitted by the following methods: obtaining information on the ship's main engine speed, propeller speed, generator load rate, and ship position; when the main engine speed is lower than a set percentage of the rated speed, the propeller speed is zero, and the ship position is within the port coordinate range, the ship is determined to be in a berthing condition; and / or when the main engine speed is zero, the generator load rate is lower than a second set threshold, and the ship position remains stationary for more than a set duration, the ship is determined to be in an anchoring condition. The normal range is determined as follows: extract the ratio of the change in capacitance to ground to the length of cables of the same type and specification, remove outliers that deviate from the mean by more than a set multiple, calculate the mean and standard deviation of the remaining ratios, and take the interval between the mean and three times the standard deviation as the normal range.

4. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that: The fault type is determined based on the combined characteristics of phase difference and harmonic distortion rate. When the phase difference is between 90 degrees capacitive lead and a set first angle capacitive lead and the harmonic distortion rate is lower than a set first threshold, it is determined to be mild insulation degradation. When the phase difference is between the first angle of capacitive lead setting and the second angle of capacitive lead setting and the harmonic distortion rate is lower than the second threshold setting, it is judged as moderate insulation degradation. When the phase difference is between the capacitive lead setting second angle and the in-phase and the harmonic distortion rate is higher than the setting second threshold, it is judged as insulation breakdown; When the leakage current exhibits intermittent pulse changes and the frequency of these changes exceeds a set frequency threshold, it is identified as an intermittent fault.

5. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that: The frequency of the high-frequency carrier signal is determined based on the cable length and the expected positioning accuracy: the total cable length of the faulty branch is obtained, the minimum frequency required is calculated based on the propagation speed of the high-frequency signal in the cable and the required positioning accuracy, and a frequency that avoids the interference band of ship power is selected as the carrier frequency based on the minimum frequency. The selection of multiple accessible locations along the cable includes setting measurement positions at the cable head, branch junction box locations, cable tray seals, cable support fixing points, and cable ends. The interval between measurement positions is determined according to the required positioning accuracy.

6. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that, Also includes: After determining that an insulation degradation fault exists, the amplitude change trend of the neutral point to ground voltage is continuously monitored. The change trend of the neutral point to ground voltage and the change trend of the active component of the leakage current are analyzed in a time series correlation. When the neutral point to ground voltage increases monotonically and shows a synchronous growth trend with the active component of the leakage current, it is used as an auxiliary confirmation basis for the insulation degradation fault.

7. The automatic detection method for online fault diagnosis of marine cables according to claim 1, characterized in that, Also includes: Ambient temperature data were collected at multiple locations along the cable to establish a statistical relationship model between the active component of leakage current and ambient temperature. Based on the current ambient temperature, the expected leakage current value based on the temperature effect was calculated. The actual measured active component of leakage current was compared with the expected value. When the actual value exceeded the set multiple of the expected value, the excess part was determined to be abnormal leakage caused by insulation deterioration. The vibration acceleration signal of the ship's main engine is collected, the main frequency and amplitude of the vibration are extracted, and the time-domain fluctuation of leakage current and / or ground capacitance is analyzed with the vibration signal in the frequency domain. When the fluctuation frequency of leakage current and / or ground capacitance is found to be consistent with the main frequency of vibration and the correlation coefficient exceeds the set threshold, it is determined to be an intermittent fault induced by vibration.

Images