A detection method for ship power batteries based on fault type recognition

Abstract

The present invention discloses a method for detecting marine power batteries based on fault type identification, which relates to the field of power battery technology. The method comprises: performing preliminary inspections of the battery compartment, battery module, electrical system, and battery management system of a marine power battery; collecting voltage, current, and temperature characterization parameters of the marine power battery cells to establish a voltage matrix for the battery cells; performing two-dimensional clustering of the matrix using a density clustering method, and determining whether a marine power battery cell is faulty based on the clustering results; constructing a directed graph model; using a data-driven method to identify module fault types; taking appropriate treatment measures and generating a test report; and predicting the degradation trend of the battery pack through big data analysis. By performing two-dimensional clustering of the matrix using a density clustering method, the influence of data noise can be reduced, thereby more accurately identifying faulty battery cells.

Description

Technical Field

[0001] The present invention relates to the technical field of power batteries, and in particular to a method for detecting power batteries for ships based on fault type identification. Background Art

[0002] With the rapid development of the global shipping industry, the performance and safety of ships, as important means of water transportation, are receiving increasing attention. Marine power batteries are core components of ship electric propulsion systems, and their stability and reliability are directly related to the operational efficiency and safety of ships.

[0003] However, due to the complex and ever-changing operating environment of ships, power battery systems are inevitably subject to various faults during use, such as battery cell performance degradation, electrical system short circuits, and battery management system failures. These faults not only affect the normal operation of the ship, but may also pose a threat to the safety of crew members and cargo. Summary of the Invention

[0004] In order to solve the above technical problems, a method for detecting marine power batteries based on fault type identification is provided. This technical solution solves the problems raised in the above background technology.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A method for detecting a marine power battery based on fault type identification, comprising:

[0007] Conduct preliminary inspections on the battery compartment, battery module, electrical system, and battery management system of marine power batteries;

[0008] Collecting voltage, current and temperature characterization parameters of the ship's power battery cells and establishing a voltage simultaneous matrix of the battery cells, wherein the voltage simultaneous matrix includes a voltage matrix, a median matrix, a voltage deviation matrix and a voltage offset matrix;

[0009] Performing two-dimensional clustering on the simultaneous matrix using a density clustering method, and determining whether a single battery cell for a ship power battery is faulty based on the clustering result;

[0010] Based on the fault propagation relationship of the ship power battery system, a directed graph model is constructed;

[0011] The neural network fault location method is used to locate the module where the fault occurs. At the module-component layer, a data-driven method is used to identify the module fault type.

[0012] Based on the fault type, take appropriate treatment measures and generate a test report, which includes infrared thermal images, insulation test values, voltage consistency data, fault type, and treatment measures;

[0013] The data from each inspection is entered into the ship health management system, and the degradation trend of the battery pack is predicted through big data analysis.

[0014] Preferably, the collecting of voltage, current and temperature characterization parameters of the ship power battery cells and establishing a voltage simultaneous matrix of the battery cells specifically includes:

[0015] Use the number to uniquely identify each battery cell, and connect the input channel of the data acquisition system to the positive and negative poles, current sensor and temperature sensor of the battery cell;

[0016] Based on the voltmeter, current sensor and temperature sensor, the data acquisition system is started to collect the voltage, current and temperature data of each battery cell in real time;

[0017] Based on the operating characteristics and requirements of the battery system, set the sampling interval and filter the collected raw data to remove noise and outliers;

[0018] Perform time synchronization processing on the filtered data to correspond the voltage, current and temperature data at the same time point;

[0019] The collected voltage data is sorted based on the battery cell number and sampling time point to form a two-dimensional voltage matrix;

[0020] The rows of the voltage matrix are the numbers of the battery cells, the columns of the voltage matrix are the sampling time points, and the elements in the voltage matrix are the voltage values of the battery cells with corresponding numbers at the corresponding sampling time points;

[0021] Calculate the median of the voltage data at each sampling time point, record and output the median voltage at each time point;

[0022] Arranging the voltage medians based on the sampling time points to form a median matrix reflecting the overall voltage levels of the battery pack at different time points;

[0023] Calculate the deviation between the voltage of each battery cell at each sampling time point and the median voltage at the corresponding time point, and organize the deviation values based on the battery cell number and sampling time point to form a voltage deviation matrix to reflect the inconsistency of voltage between battery cells;

[0024] For each battery cell, calculate the difference in voltage values at adjacent time points to obtain its voltage change at each sampling time point;

[0025] Arrange the voltage variation based on the battery cell number and the sampling time point to form a voltage offset matrix, reflecting the change trend of the battery cell voltage over time;

[0026] The voltage matrix, median matrix, voltage deviation matrix and voltage offset matrix are jointly analyzed to form a comprehensive battery cell voltage status assessment system.

[0027] Preferably, the performing two-dimensional clustering on the simultaneous matrix by using a density clustering method, and determining whether a single ship power battery cell has a fault based on the clustering result specifically includes:

[0028] Based on the distribution characteristics of the simultaneous matrix data, the neighborhood radius parameter of the density clustering algorithm is determined by drawing the k-distance graph;

[0029] Based on the number of noise points in the data set, the minimum number of points parameter of the density clustering algorithm is preset to be greater than the number of noise points in the data set;

[0030] The voltage deviation matrix and the voltage offset matrix were selected from the simultaneous matrix as clustering dimensions for two-dimensional cluster analysis;

[0031] Input the selected two-dimensional data into a density clustering algorithm, perform cluster analysis, and output clustering results, wherein the clustering results include core points, boundary points, and noise points;

[0032] Use the scatter plot visualization tool to display the clustering results. Each cluster is represented by a different color, and the noise points are distinguished by markers.

[0033] Based on the clustering results, analyze the characteristics of each cluster;

[0034] Determine whether the number of battery cells in each cluster is less than a preset number and the voltage deviation or voltage offset is greater than a preset threshold. If so, output that the battery cell is faulty; if not, no output is made;

[0035] Battery cells that are confirmed to have faults are marked, and the clustering parameter setting process is optimized based on the clustering results and fault judgment history.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The density clustering method is used to perform two-dimensional clustering of the simultaneous matrix, which can reduce the impact of data noise and thus more accurately identify faulty battery cells. It has strong system compatibility and adaptability and can be applied to ship power battery systems of different types or brands without a large number of targeted adjustments and optimizations. It predicts the degradation trend of the battery pack based on big data analysis, realizes real-time monitoring and early warning of the ship power battery system, simplifies the maintenance process of the ship power battery system through automation and intelligent means, reduces the need for manual intervention and regular inspections, and reduces maintenance costs and time costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1This is a flow chart of a method for detecting a marine power battery based on fault type identification according to the present invention;

[0039] Figure 2 This is a flow chart of a method for performing preliminary inspections on the battery compartment, battery module, electrical system, and battery management system of a marine power battery according to the present invention;

[0040] Figure 3 A flow chart of the method for establishing a voltage simultaneous matrix of battery cells according to the present invention;

[0041] Figure 4 This is a flow chart of a method for performing two-dimensional clustering on the simultaneous matrix using a density clustering method according to the present invention;

[0042] Figure 5 A flow chart of the method for constructing a directed graph model of the present invention;

[0043] Figure 6 This is a flow chart of a method for identifying module fault types using a data-driven approach according to the present invention;

[0044] Figure 7 This is a flow chart of the method for predicting the degradation trend of a battery pack through big data analysis according to the present invention. DETAILED DESCRIPTION

[0045] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0046] Reference Figure 1 As shown, a method for detecting a power battery for a ship based on fault type identification includes:

[0047] Conduct preliminary inspections on the battery compartment, battery module, electrical system, and battery management system of marine power batteries;

[0048] Collecting voltage, current and temperature characterization parameters of the ship's power battery cells and establishing a voltage simultaneous matrix of the battery cells, wherein the voltage simultaneous matrix includes a voltage matrix, a median matrix, a voltage deviation matrix and a voltage offset matrix;

[0049] Performing two-dimensional clustering on the simultaneous matrix using a density clustering method, and determining whether a single battery cell for a ship power battery is faulty based on the clustering result;

[0050] Based on the fault propagation relationship of the ship power battery system, a directed graph model is constructed;

[0051] The neural network fault location method is used to locate the module where the fault occurs. At the module-component layer, a data-driven method is used to identify the module fault type.

[0052] Based on the fault type, take appropriate treatment measures and generate a test report, which includes infrared thermal images, insulation test values, voltage consistency data, fault type, and treatment measures;

[0053] The data from each inspection is entered into the ship health management system, and the degradation trend of the battery pack is predicted through big data analysis.

[0054] Reference Figure 2 As shown, preliminary inspections of the battery compartment, battery module, electrical system and battery management system of the marine power battery are carried out separately, specifically including:

[0055] Use temperature and humidity sensors to detect the temperature and humidity inside the battery compartment, check the operating status of the ventilation system inside the compartment, and reduce heat accumulation generated during battery operation;

[0056] Use a smoke generator to test the tightness of the cabin, evaluate the effectiveness of the electromagnetic shielding measures in the cabin, and reduce the impact of electromagnetic interference on the battery management system and other electronic equipment on board;

[0057] Use a ground resistance tester to check the grounding system in the battery compartment to reduce static electricity accumulation and lightning damage;

[0058] Check the firmness of the connection cables between modules and between modules and the battery management system;

[0059] Measure the voltage of each battery cell and calculate the voltage deviation between each battery cell to maintain voltage balance and avoid overcharging and over-discharging of cells;

[0060] Use an internal resistance tester to measure the internal resistance of the battery module, analyze the internal resistance change trend, and predict the battery health status;

[0061] Observe and record the temperature distribution of each part of the module and detect temperature uniformity;

[0062] Use a megohmmeter to test the insulation resistance of electrical circuits and detect the insulation performance between circuits and between circuits and ground;

[0063] Simulate overload conditions to verify the response speed of the overload protection device and conduct short-circuit tests to check the effectiveness of the short-circuit protection mechanism;

[0064] Check the communication links between the battery management system and the battery modules and the onboard monitoring system, and verify the compatibility of the battery management system communication protocol with other systems on board;

[0065] Simulate fault conditions to verify the speed at which the battery management system can identify the fault type and issue an early warning signal.

[0066] Ensure that the temperature and humidity in the cabin are within the range recommended by the battery manufacturer to reduce heat accumulation generated by the battery during operation and maintain battery performance. Ensure that the ventilation system can effectively discharge the heat in the cabin and maintain a suitable environment in the cabin. Evaluate the effectiveness of the electromagnetic shielding measures in the cabin. This can be done with professional electromagnetic shielding test equipment to ensure that the impact of electromagnetic interference on the battery management system and other electronic equipment on board is minimized. Ensure that the grounding system is in good condition to reduce the risk of static electricity accumulation and lightning damage.

[0067] Reference Figure 3 As shown in the figure, the voltage, current and temperature characterization parameters of the ship power battery monomer are collected to establish the voltage simultaneous matrix of the battery monomer, which specifically includes:

[0068] Use the number to uniquely identify each battery cell, and connect the input channel of the data acquisition system to the positive and negative poles, current sensor and temperature sensor of the battery cell;

[0069] Based on the voltmeter, current sensor and temperature sensor, the data acquisition system is started to collect the voltage, current and temperature data of each battery cell in real time;

[0070] Based on the operating characteristics and requirements of the battery system, set the sampling interval and filter the collected raw data to remove noise and outliers;

[0071] Perform time synchronization processing on the filtered data to correspond the voltage, current and temperature data at the same time point;

[0072] The collected voltage data is sorted based on the battery cell number and sampling time point to form a two-dimensional voltage matrix;

[0073] The rows of the voltage matrix are the numbers of the battery cells, the columns of the voltage matrix are the sampling time points, and the elements in the voltage matrix are the voltage values of the battery cells with corresponding numbers at the corresponding sampling time points;

[0074] Calculate the median of the voltage data at each sampling time point, record and output the median voltage at each time point;

[0075] Arranging the voltage medians based on the sampling time points to form a median matrix reflecting the overall voltage levels of the battery pack at different time points;

[0076] Calculate the deviation between the voltage of each battery cell at each sampling time point and the median voltage at the corresponding time point, and organize the deviation values based on the battery cell number and sampling time point to form a voltage deviation matrix to reflect the inconsistency of voltage between battery cells;

[0077] For each battery cell, calculate the difference in voltage values at adjacent time points to obtain its voltage change at each sampling time point;

[0078] Arrange the voltage variation based on the battery cell number and the sampling time point to form a voltage offset matrix, reflecting the change trend of the battery cell voltage over time;

[0079] The voltage matrix, median matrix, voltage deviation matrix and voltage offset matrix are jointly analyzed to form a comprehensive battery cell voltage status assessment system.

[0080] Connect the input channel of the data acquisition system to the positive and negative electrodes of the battery cell to collect voltage data; connect the current sensor to the current output end of the battery cell to collect current data; connect the temperature sensor to the surface or inside of the battery cell to collect temperature data.

[0081] Reference Figure 4 As shown, the density clustering method is used to perform two-dimensional clustering on the simultaneous matrix, and based on the clustering results, it is determined whether there is a fault in the marine power battery cell. Specifically, the method includes:

[0082] Based on the distribution characteristics of the simultaneous matrix data, the neighborhood radius parameter of the density clustering algorithm is determined by drawing the k-distance graph;

[0083] Based on the number of noise points in the data set, the minimum number of points parameter of the density clustering algorithm is preset to be greater than the number of noise points in the data set;

[0084] The voltage deviation matrix and the voltage offset matrix were selected from the simultaneous matrix as clustering dimensions for two-dimensional cluster analysis;

[0085] Input the selected two-dimensional data into a density clustering algorithm, perform cluster analysis, and output clustering results, wherein the clustering results include core points, boundary points, and noise points;

[0086] Use the scatter plot visualization tool to display the clustering results. Each cluster is represented by a different color, and the noise points are distinguished by markers.

[0087] Based on the clustering results, analyze the characteristics of each cluster;

[0088] Determine whether the number of battery cells in each cluster is less than a preset number and the voltage deviation or voltage offset is greater than a preset threshold. If so, output that the battery cell is faulty; if not, no output is made;

[0089] Battery cells that are confirmed to have faults are marked, and the clustering parameter setting process is optimized based on the clustering results and fault judgment history.

[0090] Based on the distribution characteristics of the simultaneous matrix data, the neighborhood radius parameter of the density clustering algorithm is determined by drawing a k-distance graph. The k-distance graph shows the distance distribution from each point in the data set to its kth nearest neighbor. By observing the "inflection point" of the k-distance graph, that is, the point where the distance suddenly increases, a suitable neighborhood radius parameter value can be determined, which can distinguish high-density areas from low-density areas. Usually, the value of the minimum number of points parameter can be determined through domain knowledge or preliminary experiments. If the noise points in the data set are expected to not exceed 5%, the minimum number of points parameter can be set to an integer greater than 5% of the data set size. It is judged whether the number of battery cells in each cluster is less than a preset number, such as 5, and whether the voltage deviation or voltage offset is greater than a preset threshold, such as the voltage deviation exceeds ±5%, and the voltage offset exceeds ±10mV.

[0091] Reference Figure 5 As shown in the figure, based on the fault propagation relationship of the ship power battery system, the directed graph model is constructed, which specifically includes:

[0092] Divide the ship's power battery system into at least one component, with each component as a node. The battery compartment, battery module, electrical system, and battery management system are defined as different nodes.

[0093] Identify fault propagation paths between components based on system operating principles and historical fault data;

[0094] Based on the fault propagation path, a directed edge is established in the directed graph. The directed edge points from the fault source component to the affected component, indicating the propagation direction of the fault.

[0095] Based on the sorted components and fault propagation relationships, a directed graph model is initially constructed;

[0096] By merging similar nodes and simplifying the fault propagation path, the initially constructed directed graph model is optimized to obtain the final directed graph model.

[0097] Merge similar nodes: Check the initially constructed directed graph model, identify and merge nodes with similar functions or consistent fault impacts to reduce model complexity; Simplify fault propagation paths: Analyze the fault propagation paths in the directed graph, delete redundant or less influential paths, and retain only the main fault propagation paths; Based on the optimized model structure, re-evaluate and adjust the weights of directed edges to ensure that they can more accurately reflect the probability and severity of fault propagation.

[0098] Reference Figure 6 As shown in the figure, the neural network fault location method is used to locate the module where the fault occurs. At the module-component layer, a data-driven method is used to identify the module fault type, specifically including:

[0099] Use historical fault data and corresponding fault modules to train the neural network model and learn the mapping relationship between fault features and fault modules;

[0100] During the training process, a cross-validation strategy is used to prevent overfitting and improve the generalization ability of the model;

[0101] The pre-processed real-time operation data is input into the trained neural network model, and the model outputs the module with the highest failure probability as the suspected fault module;

[0102] For suspected fault modules, further extracting their detailed operation data features, wherein the detailed operation data features include waveform features and statistical features;

[0103] The recognition model is trained using historical fault data and corresponding fault types. The extracted operating data features of the suspected fault module are input into the trained recognition model, and the model outputs the fault type with the highest probability as the recognition result.

[0104] Extract fault features from historical fault data. These features can be sensor readings, operating parameters, environmental conditions, etc., depending on the system type. Standardize or normalize the feature data to ensure that different features are on the same scale, which helps the learning efficiency of the neural network.

[0105] Reference Figure 7 As shown, each test data is entered into the ship health management system, and the degradation trend of the battery pack is predicted through big data analysis, including:

[0106] Integrate each test data with other test data in the ship health management system to form a complete test data set;

[0107] Extracting characteristic variables related to the battery pack degradation trend from the integrated complete detection data set, wherein the characteristic variables include the number of charge and discharge cycles, capacity attenuation rate, and temperature change rate of the battery;

[0108] By training the model, the mapping relationship between the battery pack degradation trend and the characteristic variables is learned;

[0109] Real-time detection data is input into the trained model to predict the battery pack degradation trend. The prediction results include the remaining life of the battery pack, capacity decay rate, and potential failure point information.

[0110] Based on the prediction results, a battery pack degradation trend report is generated, wherein the battery pack degradation trend report includes the prediction results, data analysis charts, and conclusions and suggestions.

[0111] From the integrated complete detection data set, select characteristic variables that are closely related to the battery pack degradation trend. These characteristic variables should include but are not limited to: the number of battery charge and discharge cycles: reflecting the frequency of battery use and cumulative loss; capacity decay rate: directly measuring the reduction in battery capacity; temperature change rate: reflecting the temperature change of the battery working environment, which has an important impact on battery performance.

[0112] Furthermore, the present solution also proposes a computer-readable storage medium on which a computer-readable program is stored. When the computer-readable program is called, the above-mentioned ship power battery detection method based on fault type identification is executed.

[0113] It is understandable that the storage medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state disk (SSD).

[0114] In summary, the advantages of the present invention are: the density clustering method is used to perform two-dimensional clustering on the simultaneous matrix, which can reduce the influence of data noise, thereby more accurately identifying faulty battery cells, has strong system compatibility and adaptability, and can be applied to ship power battery systems of different types or brands without a large amount of targeted adjustment and optimization. It predicts the decline trend of the battery pack based on big data analysis, realizes real-time monitoring and early warning of the ship power battery system, simplifies the maintenance process of the ship power battery system through automation and intelligent means, reduces the need for manual intervention and regular inspections, and reduces maintenance costs and time costs.

[0115] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting marine power batteries based on fault type identification, characterized in that: include: Conduct preliminary inspections on the battery compartment, battery module, electrical system, and battery management system of marine power batteries; Collecting voltage, current and temperature characterization parameters of the ship's power battery cells and establishing a voltage simultaneous matrix of the battery cells, wherein the voltage simultaneous matrix includes a voltage matrix, a median matrix, a voltage deviation matrix and a voltage offset matrix; Performing two-dimensional clustering on the simultaneous matrix using a density clustering method, and determining whether a single battery cell for a ship power battery is faulty based on the clustering result; Based on the fault propagation relationship of the ship power battery system, a directed graph model is constructed; The neural network fault location method is used to locate the module where the fault occurs. At the module-component layer, a data-driven method is used to identify the module fault type. Based on the fault type, take appropriate treatment measures and generate a test report, which includes infrared thermal images, insulation test values, voltage consistency data, fault type, and treatment measures; The test data is entered into the ship health management system, and the degradation trend of the battery pack is predicted through big data analysis; The collecting of voltage, current and temperature characterization parameters of the ship power battery cells and establishing a voltage simultaneous matrix of the battery cells specifically includes: Use the number to uniquely identify each battery cell, and connect the input channel of the data acquisition system to the positive and negative poles, current sensor and temperature sensor of the battery cell; Based on the voltmeter, current sensor and temperature sensor, the data acquisition system is started to collect the voltage, current and temperature data of each battery cell in real time; Based on the operating characteristics and requirements of the battery system, set the sampling interval and filter the collected raw data to remove noise and outliers; Perform time synchronization processing on the filtered data to correspond the voltage, current and temperature data at the same time point; The collected voltage data is sorted based on the battery cell number and sampling time point to form a two-dimensional voltage matrix; The rows of the voltage matrix are the numbers of the battery cells, the columns of the voltage matrix are the sampling time points, and the elements in the voltage matrix are the voltage values of the battery cells with corresponding numbers at the corresponding sampling time points; Calculate the median of the voltage data at each sampling time point, record and output the median voltage at each time point; Arranging the voltage medians based on the sampling time points to form a median matrix reflecting the overall voltage levels of the battery pack at different time points; Calculate the deviation between the voltage of each battery cell at each sampling time point and the median voltage at the corresponding time point, and organize the deviation values based on the battery cell number and sampling time point to form a voltage deviation matrix to reflect the inconsistency of voltage between battery cells; For each battery cell, calculate the difference in voltage values at adjacent time points to obtain its voltage change at each sampling time point; Arrange the voltage variation based on the battery cell number and the sampling time point to form a voltage offset matrix, reflecting the change trend of the battery cell voltage over time; The voltage matrix, median matrix, voltage deviation matrix and voltage offset matrix are jointly analyzed to form a comprehensive battery cell voltage status assessment system.

2. A method for detecting marine power batteries based on fault type identification according to claim 1, characterized in that: The preliminary inspection of the battery compartment, battery module, electrical system and battery management system of the marine power battery specifically includes: Use temperature and humidity sensors to detect the temperature and humidity inside the battery compartment, check the operating status of the ventilation system inside the compartment, and reduce heat accumulation generated during battery operation; Use a smoke generator to test the tightness of the cabin, evaluate the effectiveness of the electromagnetic shielding measures in the cabin, and reduce the impact of electromagnetic interference on the battery management system and other electronic equipment on board; Use a ground resistance tester to check the grounding system in the battery compartment to reduce static electricity accumulation and lightning damage; Check the firmness of the connection cables between modules and between modules and the battery management system; Measure the voltage of each battery cell and calculate the voltage deviation between each battery cell to maintain voltage balance and avoid overcharging and over-discharging of cells; Use an internal resistance tester to measure the internal resistance of the battery module, analyze the internal resistance change trend, and predict the battery health status; Observe and record the temperature distribution of each part of the module and detect temperature uniformity; Use a megohmmeter to test the insulation resistance of electrical circuits and detect the insulation performance between circuits and between circuits and ground; Simulate overload conditions to verify the response speed of the overload protection device and conduct short-circuit tests to check the effectiveness of the short-circuit protection mechanism; Check the communication links between the battery management system and the battery modules and the onboard monitoring system, and verify the compatibility of the battery management system communication protocol with other systems on board; Simulate fault conditions to verify the speed at which the battery management system can identify the fault type and issue an early warning signal.

3. A method for detecting marine power batteries based on fault type identification according to claim 2, characterized in that: The method of performing two-dimensional clustering on the simultaneous matrix using a density clustering method and determining whether a single ship power battery cell has a fault based on the clustering result specifically includes: Based on the distribution characteristics of the simultaneous matrix data, the neighborhood radius parameter of the density clustering algorithm is determined by drawing the k-distance graph; Based on the number of noise points in the data set, the minimum number of points parameter of the density clustering algorithm is preset to be greater than the number of noise points in the data set; The voltage deviation matrix and the voltage offset matrix were selected from the simultaneous matrix as clustering dimensions for two-dimensional cluster analysis; Input the selected two-dimensional data into a density clustering algorithm, perform cluster analysis, and output clustering results, wherein the clustering results include core points, boundary points, and noise points; Use the scatter plot visualization tool to display the clustering results. Each cluster is represented by a different color, and the noise points are distinguished by markers. Based on the clustering results, analyze the characteristics of each cluster; Determine whether the number of battery cells in each cluster is less than a preset number and the voltage deviation or voltage offset is greater than a preset threshold. If so, output that the battery cell is faulty; if not, no output is made; Battery cells that are confirmed to have faults are marked, and the clustering parameter setting process is optimized based on the clustering results and fault judgment history.

4. A method for detecting marine power batteries based on fault type identification according to claim 3, characterized in that: The construction of the directed graph model based on the fault propagation relationship of the ship power battery system specifically includes: Divide the ship's power battery system into at least one component, with each component as a node. The battery compartment, battery module, electrical system, and battery management system are defined as different nodes. Identify fault propagation paths between components based on system operating principles and historical fault data; Based on the fault propagation path, a directed edge is established in the directed graph. The directed edge points from the fault source component to the affected component, indicating the propagation direction of the fault. Based on the sorted components and fault propagation relationships, a directed graph model is initially constructed; By merging similar nodes and simplifying the fault propagation path, the initially constructed directed graph model is optimized to obtain the final directed graph model.

5. A method for detecting marine power batteries based on fault type identification according to claim 4, characterized in that: The neural network fault location method is used to locate the module where the fault occurs. At the module-component layer, a data-driven method is used to identify the module fault type, specifically including: Use historical fault data and corresponding fault modules to train the neural network model and learn the mapping relationship between fault features and fault modules; During the training process, a cross-validation strategy is used to prevent overfitting and improve the generalization ability of the model; The pre-processed real-time operation data is input into the trained neural network model, and the model outputs the module with the highest failure probability as the suspected fault module; For suspected fault modules, further extracting their detailed operation data features, wherein the detailed operation data features include waveform features and statistical features; The recognition model is trained using historical fault data and corresponding fault types. The extracted operating data features of the suspected fault module are input into the trained recognition model, and the model outputs the fault type with the highest probability as the recognition result.

6. A method for detecting marine power batteries based on fault type identification according to claim 5, characterized in that: The recording of each test data into the ship health management system and the prediction of battery pack degradation trends through big data analysis specifically include: Integrate each test data with other test data in the ship health management system to form a complete test data set; Extracting characteristic variables related to the battery pack degradation trend from the integrated complete detection data set, wherein the characteristic variables include the number of charge and discharge cycles, capacity attenuation rate, and temperature change rate of the battery; By training the model, the mapping relationship between the battery pack degradation trend and the characteristic variables is learned; Real-time detection data is input into the trained model to predict the battery pack degradation trend. The prediction results include the remaining life of the battery pack, capacity decay rate, and potential failure point information. Based on the prediction results, a battery pack degradation trend report is generated, wherein the battery pack degradation trend report includes the prediction results, data analysis charts, and conclusions and suggestions.

Images