Multi-fault intelligent diagnosis and safety control method for joint debugging box of ship shore power system

Abstract

The invention relates to the technical field of ship electrical systems, in particular to a ship shore power system joint debugging box multi-fault intelligent diagnosis and safety control method. The method comprises the following steps: firstly, installing a plurality of sensors in a joint debugging box to collect key electrical parameters in real time to form a sample vector, and carrying out mean value and standard deviation standardization processing on data to eliminate dimensional difference; then calculating a covariance matrix by using a principal component analysis method and performing eigenvalue decomposition, and selecting the first k principal components of which the cumulative contribution rate reaches 85% or above to realize dimension reduction; aiming at a specific fault, randomly dividing a training set and a test set, calculating a direction cosine value of a principal component matrix of the specific fault, performing statistics to determine an error control upper limit and an error control lower limit, and further repeatedly establishing a fault feature direction library for each fault mode; and finally, collecting new data in real time, comparing the new data with the feature library after dimension reduction processing, judging a fault if a cosine value falls into a preset interval, and automatically triggering power failure, alarm, data recording and reset detection to realize intelligent fault diagnosis and safety control.

Description

Technical Field

[0001] The present invention relates to the technical field of ship electrical systems, and particularly to a method for intelligent diagnosis and safety control of multiple faults in a ship shore power system joint debugging box. Background Art

[0002] In recent years, with the continuous improvement of the electrification level of ships and the increasingly stringent requirements for safety in maritime transportation, traditional ship electrical systems face various faults during commissioning and operation, such as single-phase to ground short circuit, phase-to-phase short circuit, open phase, overvoltage, undervoltage, and three-phase imbalance. These faults can not only cause equipment damage and electrical fires, but also seriously shorten the equipment life and reduce the system reliability. Ships with newly installed or retrofitted power receiving facilities often expose problems of frequent and diverse electrical faults in actual applications, making the traditional diagnostic methods relying on experience judgment and simple power protection difficult to meet the requirements of real-time monitoring and accurate early warning. Therefore, there is an urgent need for a method of intelligent fault diagnosis and safety control to ensure that the ship shore power system can quickly cut off power, give an alarm, and record fault data when abnormalities occur, thereby providing scientific and effective technical support for the stable operation of the ship electrical system. Summary of the Invention

[0003] The object of the present invention is to propose a method for intelligent diagnosis and safety control of multiple faults in a ship shore power system joint debugging box in view of the problems in the background art.

[0004] The technical solution of the present invention: A method for intelligent diagnosis and safety control of multiple faults in a ship shore power system joint debugging box specifically includes the following steps:

[0005] S1. Real-time collect key electrical parameters;

[0006] S2. Standardize the collected key electrical parameters;

[0007] S3. Integrate all standardized samples into a data matrix, obtain an eigenvector matrix by calculating the covariance matrix and its eigenvalue decomposition, select the first k eigenvectors after arranging the eigenvalues in descending order, and then project the data into a low-dimensional space using the eigenvectors;

[0008] S4. Randomly divide the training set and the test set in the fault state, perform standardized preprocessing and principal component analysis on the two sets of data, calculate the cosine similarity between the principal component matrices, and determine the upper and lower limits of the fault error through multiple grouped statistics;

[0009] S5. Identify the faults occurring in the ship electrical system, analyze to obtain their respective eigenvectors and the corresponding cosine value statistical results, determine the upper and lower limits of the error for each fault mode, and organize the information of each fault mode into a fault feature direction library;

[0010] S6. Collect new samples in real time, obtain feature vectors through standardization and dimensionality reduction, conduct fault analysis, calculate the cosine similarity with each mode vector in the fault feature library, and determine it as a fault if the similarity is within the preset error range, otherwise consider it normal;

[0011] S7. When the cosine similarity exceeds the limit, the system powers off, alarms, records data, and resets automatically.

[0012] Preferably, the standardization process is as follows:

[0013] For each component of each sample, use the following formula:

[0014]

[0015] where, Z i is the i-th sample after standardization; m is the sample mean vector; n represents the total number of samples; v is the sample standard deviation vector, and its component calculation formula is:

[0016]

[0017] where, x ij represents the j-th variable in the i-th sample; m j is the mean of the j-th variable.

[0018] Preferably, the projection process of projecting data into a low-dimensional space using feature vectors is as follows:

[0019] S31. Integrate all standardized samples into a data matrix: Z = [Z1, Z2, …, Z i , …, Z n ;

[0020] where, Z represents the data matrix of all integrated standardized samples;

[0021] S32. Calculate the covariance matrix S:

[0022]

[0023] where, T represents the transpose operation of the matrix;

[0024] S33. Perform eigenvalue decomposition on the covariance matrix S to solve the eigenvalues and the corresponding eigenvectors:

[0025] SW = WΛ;

[0026] where, W is the eigenvector matrix, W = [w1,.., w i , …, w p ; w i represents any direction; Λ is a diagonal matrix, and the elements on its diagonal are λ1, λ2, …, λp Successively represent the variance contributions in each direction;

[0027] S34. Arrange in descending order according to the eigenvalue magnitudes, and select the top k eigenvectors W k =[w1, w2, …, w k such that the cumulative contribution rate satisfies:

[0028]

[0029] S35. Use the selected eigenvectors W k to project the data into a low-dimensional space to obtain the dimensionality-reduced data Y:

[0030]

[0031] Preferably, the generation process of the upper and lower limits of the fault error is as follows:

[0032] S41. Extract the samples in the fault state from the collected data, and then randomly divide them into a training set X train and a test set X test ;

[0033] S42. Generate the principal component matrices W train and W test ;

[0034] S43. Calculate the direction cosine values:

[0035]

[0036] where cos<W train , W test > represents the direction cosine value of the principal component matrices W train and W test ; W train ·W test is the dot product between the corresponding eigenvectors; ||W train || and ||W test || are the Euclidean norms of the vectors in each matrix respectively;

[0037] S44. Randomly group the samples multiple times to calculate the cosine values, and statistically determine the upper and lower limits of the error for the fault state.

[0038] Preferably, the construction process of the fault feature direction library is as follows: Define the faults type j that occur in the ship electrical system, analyze to obtain their respective eigenvectors W j and the corresponding cosine value statistical results, determine the upper and lower limits of the error LCL j and UCL j for each fault mode, and finally organize the information of each fault mode into the fault feature direction library {typej ,W j ,LCL j ,UCL j}}。

[0039] Preferably, the fault analysis process is as follows:

[0040] S61. Obtain a new sample X new , and perform standardization and dimensionality reduction processing on it to obtain the feature vector W of the new data new ;

[0041] S62. Compare W new with the feature vectors W of each fault mode in the feature library j , and calculate the cosine similarity:

[0042] S63. If a certain cosine value falls within the error control interval [LCL j , UCL j of the corresponding fault mode, it is determined that the new data matches the fault; otherwise, it is regarded as the normal state or not reaching the fault standard.

[0043] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0044] The present invention designs a multi-fault intelligent diagnosis and safety control method for the ship shore power system joint debugging box, realizing the full-process automation from data acquisition, dimensionality reduction feature extraction, fault feature library construction to real-time intelligent diagnosis and safety control, providing an efficient, accurate and safe fault prevention and handling technology for the ship shore power system:

[0045] (1) Real-time and efficient fault diagnosis: Through multi-sensor data acquisition and data standardization processing, combined with principal component analysis (PCA) to extract key features, real-time monitoring and intelligent diagnosis of various faults in the ship shore power system (such as single-phase to ground short circuit, phase-to-phase short circuit, phase loss, overvoltage, undervoltage and three-phase unbalance faults) are realized, significantly improving the accuracy of fault identification;

[0046] (2) Dimensionality reduction and noise reduction, extracting core information: Using the PCA method to perform dimensionality reduction processing on the data, reducing redundant information and suppressing noise interference, making the subsequent fault feature comparison more reliable and providing a unified standard for constructing a multi-fault feature direction library;

[0047] (3) Construction of a standard feature direction library: By performing multiple random grouping calculations on each fault mode, a standard feature direction and error control range are established, providing a strong basis for accurate comparison under different fault states and enhancing the system's ability to identify fault types;

[0048] (4) Intelligent safety control measures: After detecting a fault, automatically trigger safety measures such as power-off, alarm, data recording, and reset detection, timely isolate the fault, and prevent the spread of accidents, thereby improving the safety and reliability of the overall system;

[0049] (5) Reduce maintenance costs and risks: By implementing automated monitoring and fault warning, not only the need for manual intervention is reduced, but also the risk of equipment damage or fire accidents caused by faults is lowered, which helps to extend the equipment life and reduce maintenance costs. Brief Description of the Drawings

[0050] Figure 1 is a topology reference diagram of the existing onshore power retrofit system;

[0051] Figure 2 is the general assembly drawing of the device in the embodiment of the present invention;

[0052] Figure 3 is the electrical schematic diagram of the device in the embodiment of the present invention;

[0053] Figure 4 is the floor plan layout of the device in the embodiment of the present invention;

[0054] Figure 5 is the equivalent circuit diagram of the under-voltage control in the embodiment of the present invention;

[0055] Figure 6 is the turn-off voltage coefficient K MX characteristic analysis diagram;

[0056] Figure 7 is the under-voltage control circuit diagram in the embodiment of the present invention;

[0057] Figure 8 is the simulation diagram of the under-voltage coil voltage characteristics in the embodiment of the present invention;

[0058] Figure 9 is the construction flow chart of the fault feature direction library proposed by the present invention;

[0059] Figure 10 is the diagnostic flow chart after the debugging of the joint debugging box starts in the present invention.

[0060] Reference Signs: 1, circuit breaker; 2, primary input terminal; 3, power indicator light; 4, output live indicator light; 5, emergency stop button; 6, electric operating mechanism; 7, sampling control unit; 8, dissipating resistor; 9, first standard socket; 10, second standard socket; 11, portable handle; 101, opening; 12, three-phase power inlet hole; 13, wire tying bracket; 14, grounding post. Detailed Embodiments

[0061] Embodiment 1, A common onshore power system for transportation ships is asFigure 1 As shown in the figure, in the entire shore power system, the input is the onshore power supply pile (within the red frame), and the receiving facility of the ship shore power system (within the blue frame), which includes: 1 marine cable winch (100m specified in the propulsion plan), 2 shore power box, 3 measurement box, 4 isolation transformer; the output is connected to the ship's main switchboard. The basic functions of the receiving facility of the ship shore power system are: three-phase power supply, and emergency stop is available in case of failure. After the receiving facility of the ship shore power system is installed or modified on the ship, it is necessary to verify whether the receiving facility of the ship shore power system is installed and debugged qualified. The most ideal situation is to directly use the standard power supply pile facility on the shore for debugging and inspection, but this ideal situation is restricted by external conditions: 1 The receiving facility that has not been tested and verified is not allowed to be connected to the standard power supply pile; 2 Whether the ship is at the berth; 3 Whether there is a standard power supply pile on the shore; 4 Whether the time period for ship test debugging occupies the normal power supply operation time period of the wharf, etc. These inconveniences have hindered the development of the commissioning and acceptance work of the receiving facility of the shore power system on the ship.

[0062] In addition, generally, the larger the gross tonnage of the ship, the larger the rated capacity of the load, which is mainly determined by the size of the ship's load. The shore power system interfaces of most ships in inland rivers and the Yangtze River are 63A and 125A. If the ship's receiving facility with 63A requires a joint debugging system with a 63A standard interface for debugging and inspection, similarly, the ship's receiving facility with 125A requires a joint debugging system with a 125A standard interface for debugging and inspection; according to the specification requirements, regardless of whether it is a 63A or 125A shore power system, the marine shore power cable containing the emergency stop control line and power wire is installed on the cable winch to supply power to the ship. When the ship is at the shore, the winch releases the cable and connects it to the shore power pile; when the ship leaves the port, the winch winds up the cable. The power supply cable is relatively long (to ensure that it can be connected to the onshore system, and the cable length on the winch is generally at least 100 meters), and the influence of distributed capacitance needs to be considered specially, because the distributed capacitance can be ignored at short distances, and attention must be paid in the design and application links.

[0063] As Figure 1 shown, it is a topology reference diagram of the shore power transformation system, which details the structural composition of the shore power system, including the onshore power supply pile, which provides 63A and 125A standard sockets; the marine cable winch on the ship, which connects the onshore power supply pile and the onshore power box on the ship; in this embodiment, a shore power joint debugging box is provided to play the role of replacing the onshore power supply pile within the red frame in the shore power system debugging work;

[0064] The general assembly drawing of the shore power joint debugging box is as Figure 2 shown, including a box body; the first standard socket 9 and the second standard socket 10 are respectively arranged on both sides of the box body for connecting different types of ships, and it is connected to the ship shore power system as a joint debugging tool; among them, the first standard socket 9 is a 125A marine socket; the second standard socket 10 is a 63A marine socket;

[0065] A portable handle 11 is provided at the top of the box body, and the box body can be moved to a specified position through the portable handle 11;

[0066] A circuit breaker 1, a primary input terminal 2, an electric operating mechanism 6 (i.e., an electric operating mechanism), a sampling control unit 7, and a dissipating resistor 8 are arranged inside the box body; a control circuit for controlling and switching the socket is formed by connecting and combining the circuit breaker 1, the primary input terminal 2, the electric operating mechanism 6, the sampling control unit 7, and the dissipating resistor 8;

[0067] A sealed door is hinged to the front end of the box body; a power indicator light 3, an output live indicator light 4, and an emergency stop button 5 are all integrated on the sealed door of the box body; the circuit breaker 1, the primary input terminal 2, the power indicator light 3, the output live indicator light 4, the emergency stop button 5, the electric operating mechanism 6, the sampling control unit 7, and the dissipating resistor 8 are electrically connected.

[0068] The operating handle of the circuit breaker 1 passes through the opening 101 on the front door of the box body to facilitate the opening and closing of the circuit breaker 1 outside the box; a three-phase power supply inlet hole 12 is opened at the bottom of the box body for connecting an external three-phase power supply. A separate grounding post 14 is arranged at the bottom of the box body. Specifically, in order to ensure the operation safety of the joint debugging box, an external protective ground PE is connected to the primary input terminal 2 during debugging, and the protective ground PE is connected to the box body shell and the front box door. In addition, the protective ground PE is also connected to the 125A marine standard socket 9 and the 63A marine standard socket 10.

[0069] In this embodiment, a wire tying bracket 13 for fixing the power cord is arranged along one side of the three-phase power supply inlet hole 12 to assist in fixing the power cord and prevent it from shifting during use.

[0070] As Figure 3 shown, the external three-phase power supply ABC and PE are connected to the inlet of the primary input terminal DZ1 through a primary cable; the outlet of the primary input terminal DZ1 is connected to the upper ports L1, L2, L3 of the circuit breaker 1QF through a cable; the PE is connected to the PE connection points of the 125A and 63A standard marine sockets and the separate grounding post arranged at the bottom of the box body through a cable; the three lower ports L11, L12, L13 of the circuit breaker 1QF are simultaneously connected to the three-phase interfaces in the 125A marine standard socket and the 63A marine standard socket through wires; the lines A610 and C610 are led out from the lower ports L11 and L13 of the circuit breaker 1QF, and the output indicator light Hr is connected across the lines A610 and C610; the fuses FU1 and FU2 are respectively connected in series to the upper ports L1 and L3 of the circuit breaker 1QF, and the lines A600 and C600 are led out; the power indicator light Hw is connected in parallel across the lines A600 and C600.

[0071] The A600 and C600 lines are connected in parallel and cross - connect another line; an emergency stop button SB is connected between A600 and 101; a 125A relay switch 1DK is connected between 101 and 103, and the control signal connection points P2 and P1 connected to a 125A standard marine socket are led out; a 63A relay switch 2DK is connected between 103 and 105, and the control signal connection points P2 and P1 connected to a 63A standard marine socket are led out; the 105 line is connected to the C600 line through an under - voltage release coil 1QF, and a wire is led out from the lower end of the coil and connected to a power resistor to access the control relay switch group 3DK. The sampling control unit 7 selects and outputs to control the relay group 3DK to access different numbers of dissipative resistors R1, R2, R3, R4 (same specification 50kΩ, 10W) to the upper end of the under - voltage release coil 1QF to complete the loop, and controls the heating power and reliable under - voltage release by controlling the number of dissipative resistors connected to the circuit.

[0072] In Figure 2 the joint - debugging box, there are 125A marine standard sockets and 63A marine standard sockets, which are connected to the corresponding standard sockets on the 125A and 63A shore - power boxes below in Figure 3 through a 100 - meter - long standard shore - power marine cable. Taking the debugging of the 125A marine shore - power system equipment as an example, after the cable is connected, it can be seen that the positions of the control signal connection points P2 and P1 originally led out from 101 and 103 to the standard marine socket are conducted to P2 and P1 of the ship's shore - power box socket after a 100 - meter - long cable, and an emergency stop button SB1 of the shore - power box is connected in series inside the ship's shore - power box; the 63A debugging connection is similar to the above.

[0073] The technical solution corresponding to the emergency stop button SB of this device Figure 2 In the joint - debugging box, the emergency stop button 5, SB1 corresponds to the emergency stop button of the 125A shore - power box, and SB2 corresponds to the emergency stop button of the 63A shore - power box. For different specifications of 63A and 125A ship's shore - power systems, select the appropriate test scheme for the corresponding standard interface. During debugging, this joint - debugging box can provide safe access to the test power supply for the power - receiving facilities to be tested and can realize the cut - off function of the debugging box and the ship's shore - power box.

[0074] In terms of debugging functions, this joint - debugging box can respectively realize the debugging functions of the 63A loop and the 125A loop.

[0075] The sampling control unit 7 will detect the limit detection switches equipped in the 125A marine standard socket 1 and the standard socket 2; when detecting the insertion and connection of the 125A socket, the relay switch 2DK will be automatically closed and the relay switch 1DK will be disconnected. At this time, whether it is the emergency stop button of the joint debugging box or the emergency stop button of the shore power box, both can achieve the task of cutting off the test circuit (simulating the accidental emergency power cut-off during the actual use process); similarly, when detecting the connection of the 63A socket, the relay switch 2DK will be automatically disconnected and the relay switch 1DK will be closed. The specific debugging method is as follows: In the debugging of the 63A circuit, the relay switch 1DK is closed and the relay switch 2DK is disconnected. In the current circuit state, Figure 3 between A600 and C600, the emergency stop button SB of the joint debugging box and the emergency stop button SB2 of the 63A shore power box are connected in series. At this time, whether it is the emergency stop button SB of the joint debugging box or the emergency stop button SB2 of the shore power box, both can achieve the task of cutting off the test circuit (simulating the accidental emergency power cut-off during the actual use process). In the debugging of the 125A circuit, the relay switch 1DK is disconnected and the relay switch 2DK is closed. Between A600 and C600, the emergency stop button SB of the joint debugging box and the emergency stop button SB1 of the 125A shore power box are connected in series. In terms of debugging flexibility, this joint debugging box can flexibly access the dissipative resistors R1, R2, R3, and R4 with different capacities according to the characteristics of the cable material, length, etc., and minimize the influence of the equivalent capacitance of the cable to achieve reliable opening and closing of the circuit.

[0076] The three views of the plane of the joint debugging box are as Figure 4 shown: When testing the receiving facilities of different specifications, by adjusting the opening and closing of the bypass air switches 1DK and 2DK, it is possible to flexibly select for joint debugging and testing of the 63A or 125A shore power system. The joint debugging box is designed with a portable handle, which is convenient for on-board testing at any time and anywhere.

[0077] Taking the joint debugging test of the 125A shore power system as an example, the usage and operation process will be described in detail:

[0078] Preparation stage: Open-circuit the end of the cable to measure the open-circuit impedance; short-circuit the end of the cable to measure the short-circuit impedance. After completion, the sampling control unit 7 will calculate the distributed capacitance of the cable and select the output control relay switch group 3DK to access the dissipative resistor step by step according to the distributed capacitance capacity;

[0079] Debugging stage: 1. Confirm that the circuit breaker switch is in the open position and the input end is not powered on; make the correct wiring;

[0080] 2. After determining that the wiring is reliable, power on the input end. If everything is normal, the power indicator on the panel of the joint debugging box should be lit. Close the circuit breaker 1. If everything is normal, the output indicator should be lit;

[0081] 3. The input terminal accesses the main power supply A-B-C three-phase and the protective ground PE through the opening at the bottom of the joint debugging box. Taking the 125A shore power system test as an example, the output terminal is connected to a 125A standard socket, and the relay switch 2DK is closed inside the joint debugging box.

[0082] 4. Press the emergency stop SB of the joint debugging box to test the emergency stop function of the joint debugging box; after the emergency stop function is normal, power on again, and press the emergency stop SB1 of the shore power box to ensure that the emergency stop function is normal.

[0083] 5. After the emergency stop function of the shore power box is normal, power on again and conduct the power-on operation test of the ship's power receiving facilities.

[0084] 6. After the power-on operation test of the power receiving facilities is completed - cut off the power - tidy up the wiring harness.

[0085] Response to faults: 1. During the debugging stage, when the background system detects a fault, it will immediately issue an instruction to the sampling control unit 7 to control the electric operating mechanism 6 to cut off the circuit breaker 1; the background system can select an existing online monitoring system that can match it; it should be noted that the method recorded in Implementation 2 is the monitoring diagnosis and safety control method for the joint debugging box during the debugging and use process.

[0086] 2. In case of abnormal power-off, the test should be stopped immediately, and the debugging fault log should be extracted to find and eliminate the fault source; after the fault is eliminated, start the debugging.

[0087] It should be noted that a certain amount of thermal power will be generated during the use of the joint debugging box. Since this device is not equipped with an active heat dissipation device, it should be avoided from being used continuously for a long time. After the debugging work is completed correctly, the connection of this device should be disconnected in time.

[0088] Verify the performance of the above-mentioned shore power joint debugging box through the test method of the shore power joint debugging box, including the following specific steps:

[0089] S1. Conduct impedance analysis on the circuit and construct an equivalent circuit.

[0090] S2. Conduct distributed capacitance analysis based on the constructed equivalent circuit.

[0091] S3. Conduct characteristic analysis on the equivalent circuit based on the MATLAB tool.

[0092] S4. Adjust each parameter in the equivalent circuit based on the analysis results.

[0093] S5. Use the adjusted equivalent circuit for simulation and verify the accuracy of the simulation results through experiments.

[0094] Explanation of the influence of distributed capacitance on the system emergency stop function:

[0095] During the manufacturing and function testing stage of the joint debugging box, there was no Figure 3The dissipative resistor connected in parallel across the under-voltage release 1QF R , in the electrical principle Figure 3 is manifested as a dissipative resistor R being infinite. However, during actual commissioning and operation, an abnormal situation occurred where although the emergency stop button was pressed, the under-voltage release had a probability of failing to trip open, resulting in the failure of the emergency stop during the commissioning process. In response to this abnormal working condition, after re-examining the circuit characteristics and working principle, it was determined that the reason for the failure of the under-voltage release 1QF to trip open was that the long-distance power cable had a certain distributed capacitance, resulting in the voltage across the under-voltage release coil not reaching the tripping limit even when the emergency stop was pressed.

[0096] In response to the reasons obtained from the above analysis, by connecting a certain dissipative resistor in parallel with the under-voltage coil, the voltage amplitude across the under-voltage release in the circuit is reduced. To fully understand the mechanism of action of the distributed capacitance of the long-distance connecting cable, a mechanism analysis of the control circuit is carried out here.

[0097] When the cable is long, although the emergency stop is disconnected, the long-distance control cable is equivalent to a capacitor connected in series in the circuit, and the under-voltage coil can be equivalent to a resistor and an inductor. The overall equivalent circuit structure is as follows Figure 5 shown.

[0098] The equivalent circuit includes power lines L1, L3, resistors R, R MX , inductor L MX and capacitor C;

[0099] where the resistor R MX , inductor L MX constitute a series R-L circuit, and then the capacitor C is connected in series and connected to the power line L1; one end of the resistor R MX is connected to the power line L3;

[0100] The resistor R and the resistor R MX , inductor L MX constitute a parallel R-L circuit;

[0101] In this circuit, the power supply is an alternating current with a certain frequency f, and the equivalent impedance of the inductor and capacitor is affected by the power supply frequency. It is necessary to use the equivalent impedance of the inductor and capacitor for analysis.

[0102] The equivalent impedance of the capacitor and the under-voltage coil: Z MX =(R MX +jωL MX / / R, where ω = 2πf.

[0103] The voltage across the under-voltage coil: U MX =K MX (U L1 -U L3), where the coefficient U L1 , U L3 and are the voltages on L1 and L3 respectively.

[0104] Analysis shows that when the power supply frequency f is constant, the voltage U MX on the undervoltage coil is MX positively correlated with the coefficient K MX . Here, the coefficient K

[0105] Characteristic analysis: Using the MATLAB tool, comprehensively and quantitatively analyze the mathematical relationship between the coefficient K MX of the undervoltage coil turn-off voltage U MX , the power supply frequency f, the cable distributed capacitance C, and the parallel dissipation resistance R. In mathematical calculations, take the DC resistance R MX of the undervoltage coil turn-off as 100 kΩ, L MX = 50 H, and the cable length is designed to be 100 m. The relationship diagram between the coefficient K MX of the voltage U MX on the undervoltage coil and the power supply frequency f, the cable distributed capacitance C, and the parallel dissipation resistance R can be simulated.

[0106] As Figure 6 shown, on the premise that the cable distributed capacitance is constant, when the power supply frequency f ≥ 40 Hz, as the value of the parallel dissipation resistance R on the vertical axis increases (corresponding to the decrease of the dissipation power), the coefficient K MX of the turn-off voltage U MX shows an increasing trend, that is, the voltage U MX on the corresponding undervoltage coil may be too large, resulting in the failure of disconnection. Combining with an appropriate size of the dissipation resistance can reduce the coefficient K MX of the turn-off voltage U MX to a certain extent.

[0107] When the distributed capacitance C is 0.001 uF, the dissipation resistance R changes in the range of 10 kΩ - 20 kΩ, and the coefficient K MX will never be higher than 0.03. This means that the capacitance value of the distributed capacitance C is too small at this time, and there will be no abnormal phenomenon that the undervoltage tripping 1QF cannot be disconnected.

[0108] However, when the distributed capacitance C reaches 0.1 uF and the power supply frequency f = 50 Hz, it can be seen from Figure 6 that the distribution range of the coefficient K MX expands to nearly 1. At this time, the influence of the distributed capacitance on the undervoltage tripping 1QF is intensified. It can be seen from Figure 6 that only by selecting a dissipation resistance with a value less than 5 kΩ can the coefficient K MXReduce to within 0.5.

[0109] Based on the above analysis, when selecting the dissipative resistor R, the capacitance value of the distributed capacitance C of the cable needs to be fully considered. For a larger distributed capacitance, it is necessary to correspondingly increase the dissipative power and reduce the resistance value of the dissipative resistor. For the above common range of distributed capacitance per unit length of the cable: 10 pF / m to 1000 pF / m, the conclusion is that by controlling the resistance value of the power resistor within the range of 6 to 10 kΩ, the adverse effect of the distributed capacitance of the cable on undervoltage tripping can be eliminated.

[0110] It should be noted that since the dissipative resistor is always connected to the circuit during operation, there will inevitably be a certain amount of electric power consumption on it. At the same time, the above conclusion that the resistance value of the power resistor is controlled within the range of 6 - 10 kΩ can apply to the common range of cable distributed capacitance, but the electric power P = U 2 / R consumed on it can be calculated to obtain a power range of: 16 W - 26.7 W, and such power consumption is obviously not allowed. On the one hand, high power consumption means an increase in the heat dissipation requirement of the device; on the other hand, it poses higher requirements for the reliability of the dissipative resistor.

[0111] After further analysis, it is found that the distributed capacitance per unit length of the cable mostly concentrates in the range of 100 pF / m to 300 pF / m, and the capacitance value range for a 100 - meter length is 0.01 uF - 0.03 uF. The above resistance value range of 6 kΩ - 10 kΩ for the dissipative resistor is the range taken to cope with extreme situations, but the heat dissipation power range of the above dissipative resistor of 16 W - 26.7 W is obviously unreasonable in terms of the heat - generating electric power. Therefore, the following reference table for the equivalent resistance value of the dissipative resistor to be connected is set.

[0112]

[0113] Based on the above analysis, this design further improves the design by integrating the distributed capacitance detection function inside the sampling control unit 7, and applying the three - voltage method to complete the capacitance value detection of the distributed capacitance of the cable. During the test, by measuring the open - circuit impedance Z OC and short - circuit impedance Z SC , applying the formula to obtain the characteristic impedance of the cable, and further calculating the propagation constant γ:

[0114] where l is the cable length, α is the attenuation constant, and β = 2πf / v (v is the propagation speed) is the phase constant. It should be noted that in practical applications, in the case of low loss, an approximate method (α≈0) is usually used to simplify the calculation, and then the simplified equations are used to approximate the solution;

[0115] The characteristic impedance Z0, propagation constant γ, inductance L per unit length, and capacitance C per unit length have the following relationships: Based on the above formulas, the capacitance C can be calculated as follows:

[0116] C≈β / ωZ0=1 / Z0v;

[0117] Before the debugging starts, the sampling control unit can complete the detection of the distributed capacitance of the cable through the three-voltage method before the system test. According to the measured value of the distributed capacitance of the cable, the control relay 3DK is output to stepwise select and connect the power resistors (R1, R2, R3, R4) with reference to the above table to eliminate the adverse effects of the distributed capacitance.

[0118] On the basis of the above, the LTspice tool is used for experimental verification, and a voltage simulation circuit as shown in Figure 7 is built. The capacitance is selected as 0.005uF. It includes path one (R1, L1, C1) without adding a dissipation resistor; path two (R2, R21, L2, C2) with a 50kΩ dissipation resistor in parallel; path three (R3, R31, R32, L3, C3) with two 50kΩ dissipation resistors in parallel. By measuring the voltages of nodes v1, v2, v3, and v4, the voltage across the coil can be calculated as shown in Figure 8 .

[0119] In path one without a parallel resistor, it can be seen that the voltage across the under-voltage coil is significantly higher than that in path two and path three with a dissipation resistor. In contrast, paralleling a 50kΩ dissipation resistor for a distributed capacitance of 0.005uF has an obvious effect on reducing the voltage across the under-voltage coil. Paralleling 50kΩ has a stronger effect and can reduce the voltage across the under-voltage coil to within the effective range of the protection value, which is in line with the above analysis.

[0120] Embodiment 2, as shown in Figure 9 and Figure 10 , the present invention proposes a multi-fault intelligent diagnosis and safety control method for a ship shore power system joint debugging box, which is applied to the intelligent diagnosis and safety control during the debugging process of the shore power joint debugging box described in Embodiment 1. The specific implementation steps are as follows:

[0121] S1. During the debugging process of the ship shore power system joint debugging box, its key electrical parameters are collected in real time. The key electrical parameters include but are not limited to voltage, current, and power factor;

[0122] Each set of collected data forms a sample, denoted as vector X i =[x i1 ,x i2 ,…,x ij ,…,x ip T;

[0123] Among them, X i is the i-th sample; p represents the number of variables (i.e., sensor measurement items); x ij represents the j-th variable in the i-th sample;

[0124] It should be noted that it is only necessary to collect the data obtained by the sampling control unit 7 in the shore power connection adjustment box. That is to say, the system corresponding to this method is communicatively connected to the sampling control unit 7 to receive the data of the sampling control unit 7 in real time and issue instructions to the sampling control unit 7.

[0125] S2. To eliminate the influence of different dimensions and orders of magnitude, standardize the original data. For each component of each sample, use the following formula:

[0126]

[0127] Among them, Z i is the i-th sample after standardization; m is the sample mean vector; n represents the total number of samples; v is the sample standard deviation vector, and the calculation formula for its components is:

[0128]

[0129] Among them, x ij represents the j-th variable in the i-th sample; m j is the mean of the j-th variable.

[0130] S3. Principal component analysis (PCA) and feature extraction, that is, transform the original data into features that can more obviously reflect the main changes after dimensionality reduction, laying a data structure for subsequent fault mode comparison. Specifically:

[0131] S31. Integrate all standardized samples into a data matrix: Z = [Z1, Z2, …, Z i , …, Z n ;

[0132] Among them, Z represents the data matrix of all integrated standardized samples;

[0133] S32. Calculate the covariance matrix S:

[0134]

[0135] It should be noted that each element s ij in the covariance matrix S measures the linear correlation between the i-th and j-th variables;

[0136] S33. Perform eigenvalue decomposition on the covariance matrix S to solve the eigenvalues and the corresponding eigenvectors:

[0137] SW = WΛ;

[0138] Among them, \(W\) is the eigenvector matrix, \(W = [w_1,..,w i ,…,w p \); \(w i represents any direction; \(\Lambda\) is a diagonal matrix, and its diagonal elements \(\lambda_1, \lambda_2, \ldots, \lambda p represent the variance contributions of each direction in turn;

[0139] S34. Arrange in descending order of eigenvalues, and select the first \(k\) eigenvectors \(W k = [w_1,w 2, …,w k such that the cumulative contribution rate satisfies:

[0140]

[0141] Accordingly: The selected \(W k represents the dimensionality reduction basis of the main features of the original data. Using this basis, project the data into a low-dimensional space to obtain the dimensionality reduction data:

[0142]

[0143] It should be noted that this step not only reduces data redundancy but also extracts the features describing the main information of the samples, providing a unified standard for subsequent fault feature comparison.

[0144] S4. Obtaining the feature direction of a single fault mode. By grouping and statistics, the "standard" feature direction and error range of a certain fault are established. Specifically:

[0145] S41. Extract samples in the fault state from the collected data, and then randomly divide them into a training set \(X train and a test set \(X test ;

[0146] S42. Perform S1 - S3 on the two groups of data respectively to obtain their respective principal component matrices \(W train and \(W test ;

[0147] S43. To measure the consistency of the features of the two groups of data, calculate their direction cosine values:

[0148]

[0149] Among them, \(\cos\lt W train ,W test \gt\) represents the direction cosine value of the principal component matrices \(W train and \(W test ; \(W train \cdot W test is the dot product between the corresponding eigenvectors; \(\vert\vert W train \vert\vert\) and \(\vert\vert Wtest || are the Euclidean norms of the vectors in each matrix respectively;

[0150] Accordingly: the closer the cosine value is to 1, the more consistent the two groups of feature directions are; to ensure stability, the cosine value is calculated by randomly grouping the samples multiple times, and the upper and lower limits of the error control (LCL and UCL) for the fault state are statistically determined based on this;

[0151] It should be noted that this step establishes the standard feature direction under a specific fault mode and determines the allowable error range, providing a template for constructing a multi-fault feature library.

[0152] S5. As Figure 1 shown, integrate the standards of all fault modes to form a fault feature direction library for real-time discrimination, specifically:

[0153] For the possible faults type j (including but not limited to single-phase-to-ground short circuit, phase-to-phase short circuit, open phase, overvoltage and three-phase unbalance faults) in the ship electrical system, repeat the process of step S4 to obtain the respective feature vectors W j and the corresponding cosine value statistical results, and determine the upper and lower limits of the error LCL j and UCL j for each fault mode. Finally, organize the information of each fault mode into a fault feature direction library in the form of: {type j , W j , LCL j , UCL j};

[0154] S6. As Figure 2 shown, for real-time fault detection and intelligent diagnosis, use the aforementioned standards to compare with new data in real time and quickly identify the fault type, specifically:

[0155] S61. Continuously obtain new samples X new , and perform standardization and dimensionality reduction processing on it to obtain the feature vector W new of the new data;

[0156] S62. Compare W new with the feature vector W j of each fault mode in the feature library, and calculate the cosine similarity:

[0157] S63. If a certain cosine value falls within the error control interval [LCL j , UCL j of the corresponding fault mode, it is determined that the new data matches the fault; otherwise, it is regarded as a normal state or not meeting the fault standard.

[0158] S7. When the detection system finds that the cosine value of a certain fault mode exceeds the preset normal range, the following safety measures are immediately triggered:

[0159] Measure 1: Power-off control: Control the mechanical operating structure to quickly disconnect the three-phase power supply to prevent the spread of faults and equipment damage;

[0160] Measure 2: Alarm prompt: Activate the buzzer or other alarm devices to notify the operator in real time to take necessary measures;

[0161] Measure 3: Fault data recording: Store the sampled data and diagnostic results during the fault to the built-in ROM for subsequent traceability analysis and maintenance optimization;

[0162] Measure 4: Reset detection: After the fault is eliminated, the system automatically resets and enters the next detection cycle to ensure continuous monitoring.

[0163] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made without departing from the spirit of the present invention within the knowledge scope of those skilled in the art to which the present invention pertains.

Claims

1. An intelligent diagnosis and safety control method for multiple faults of a joint debugging box of a ship shore power system, characterized in that, Specifically, it includes the following steps: S1. Collect key electrical parameters in real time; S2. Standardize the collected key electrical parameters; S3. Integrate all standardized samples into a data matrix, obtain the eigenvector matrix by calculating the covariance matrix and its eigenvalue decomposition, select the first k eigenvectors after arranging the eigenvalues in descending order, and then project the data into a low-dimensional space using the eigenvectors; S4. Randomly divide the training set and the test set in the fault state, perform standardized preprocessing and principal component analysis on the two sets of data, calculate the cosine similarity between the principal component matrices, and determine the upper and lower limits of the fault error through several group statistics; S5. Identify the faults occurring in the ship electrical system, analyze to obtain their respective eigenvectors and the corresponding cosine value statistical results, determine the upper and lower limits of the error for each fault mode, and organize the information of each fault mode into a fault feature direction library; S6. Collect new samples in real time, obtain eigenvectors through standardization and dimensionality reduction, perform fault analysis, calculate the cosine similarity with each mode vector in the fault feature library, and if the similarity is within the preset error range, it is determined as a fault, otherwise it is regarded as normal; S7. When the cosine similarity exceeds the limit, the system powers off, alarms, records data and resets automatically.

2. The intelligent diagnosis and safety control method for multiple faults of the joint debugging box of a ship shore power system according to claim 1, characterized in that The standardization process is as follows: For each component of each sample, use the following formula: Among them, Z i is the i-th sample after standardization; m is the sample mean vector; n represents the total number of samples; v is the sample standard deviation vector, and the calculation formula for its components is: where x ij represents the j-th variable in the i-th sample; m j is the mean value of the j-th variable.

3. The intelligent multi-fault diagnosis and safety control method for the joint debugging box of a ship shore power system according to claim 1, wherein, The projection process of projecting the data into a low-dimensional space using the eigenvectors is as follows: S31. Integrate all standardized samples into a data matrix: Z = [Z1, Z2, …, Z i , …, Z n ; Among them, Z represents the data matrix of all integrated standardized samples; S32. Calculate the covariance matrix S: Among them, T represents the transpose operation of the matrix; S33. Perform eigenvalue decomposition on the covariance matrix S to solve the eigenvalues and the corresponding eigenvectors: SW = WΛ; Among them, W is the feature vector matrix, W = [w1,..,w i ,…,w p ; w i represents any direction; Λ is a diagonal matrix, and its diagonal elements λ1, λ2, …, λ p successively represent the variance contributions of each direction; S34. Arrange in descending order according to the eigenvalue magnitudes, and select the top k eigenvectors W k = [w1, w 2, …, w k such that the cumulative contribution rate satisfies: S35. Use the selected eigenvector W k Project the data onto a low-dimensional space to obtain the dimensionality-reduced data Y:

4. The intelligent multi-fault diagnosis and safety control method for the joint debugging box of a ship shore power system according to claim 1, characterized in that The generation process of the upper and lower limits of the fault error is as follows: S41. Extract samples in the fault state from the collected data, and then randomly divide them into a training set X train and a test set X test ; S42. Generate the principal component matrix W train and W test ; S43. Calculate the direction cosine value: where, cos<W train , W test > represents the direction cosine value between the principal component matrices W train and W test ; W train ·W test is the dot product between the corresponding eigenvectors; ||W train || and ||W test || are the Euclidean norms of the vectors in each matrix respectively; S44. Randomly group the samples several times to calculate the cosine values, and statistically determine the upper and lower limits of the error control in the fault state.

5. The intelligent diagnosis and safety control method for multiple faults of a joint debugging box of a ship shore power system according to claim 4, characterized in that, The construction process of the fault feature direction library is as follows: Define the fault types that occur in the ship electrical system j , and analyze to obtain their respective feature vectors W j and the corresponding cosine value statistical results, and determine the upper and lower limits of the error LCL j and UCL j . Finally, organize the information of each fault mode into the fault feature direction library {type j , W j , LCL j , UCL j}.

6. The intelligent multi-fault diagnosis and safety control method for the joint debugging box of a ship shore power system according to claim 5, characterized in that, The fault analysis process is as follows: S61. Obtain a new sample X new , and perform standardization and dimensionality reduction processing on it to obtain the feature vector W of the new data new ; S62. Compare W new with the feature vector W of each fault mode in the feature library j to calculate the cosine similarity: S63. If a certain cosine value falls within the error control interval [LCL j , UCL j corresponding to a fault mode, it is determined that the new data matches the fault; otherwise, it is regarded as a normal state or not meeting the fault standard.

Images