Communication base station lithium battery fault diagnosis method, device and electronic equipment

Abstract

The application discloses a communication base station lithium battery fault diagnosis method and device and electronic equipment, relates to the technical field of lithium battery monitoring and fault diagnosis, and is used for solving the problems that the communication base station lithium battery fault diagnosis is inaccurate and it is difficult to locate the fault source; the application collects multi-dimensional data through a battery management unit, including voltage, current, temperature and electrochemical impedance spectrum; based on the collected data, static and dynamic inconsistency characteristics are extracted, and a comprehensive inconsistency index is calculated; similarity matching is performed in combination with an impedance deviation spectrum and a fault feature library, a fault type and a level are identified, fault monomer space positioning, electrical topology correlation analysis, impedance spectrum depth analysis and thermal behavior anomaly analysis are performed, and the fault source is comprehensively determined; and the accuracy of fault diagnosis and source positioning is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of lithium battery monitoring and fault diagnosis technology, specifically to methods, devices, and electronic equipment for fault diagnosis of lithium batteries in communication base stations. Background Technology

[0002] With the full deployment of 5G communication networks, the in-depth advancement of the new infrastructure construction strategy, and the accelerated transformation of the communications industry towards intelligence, greenness, and high reliability, real-time monitoring of the backup power system's operational status, accurate fault diagnosis, and predictive maintenance of communication base stations, as the core infrastructure of mobile communication systems, have become crucial for ensuring network continuity and improving service quality. Simultaneously, driven by the strong demand for predictive maintenance and energy management optimization, integrating advanced technologies such as multi-dimensional data sensing and electrochemical mechanism analysis to achieve accurate assessment of battery health status and fault tracing is becoming the core path for technological upgrades in the field of intelligent operation and maintenance of communication base stations. However, existing technologies still have many problems:

[0003] Traditional technologies lack the ability to deeply integrate fault diagnosis models with multi-source, time-varying, and heterogeneous data generated at the lithium battery operation site of communication base stations. Diagnostic strategies fail to systematically integrate the collaborative constraints and physical correlations of multi-dimensional information such as real-time voltage and current waveforms, temperature field distribution, electrochemical impedance spectroscopy, historical charge and discharge records, and environmental operating parameters. Existing methods lack trend prediction and early warning mechanisms for the dynamic evolution of battery performance. Fault identification heavily relies on threshold judgment at a single moment, failing to effectively capture hidden fault modes such as gradual decay and intermittent anomalies. At the same time, diagnostic algorithms have weak modeling capabilities for complex factors such as the dynamic aggravation of inconsistencies between individual cells within the battery pack, electrical topology coupling effects, and local environmental disturbances. Fault root cause localization lacks in-depth analysis of spatial distribution characteristics and electrochemical mechanisms, making it difficult to distinguish different fault causes such as battery aging, connection degradation, and thermal management failure. Furthermore, existing technologies lack adaptability and robustness in areas such as weak feature signal extraction under low signal-to-noise ratio environments, abnormal pattern recognition under strong interference conditions, and semantic understanding of fragmented operation and maintenance data. As a result, the diagnostic system constructed is unable to meet the high standards of communication network operation and maintenance requirements in terms of accuracy, reliability, and practicality when facing complex operating conditions such as drastic changes in coverage temperature, large-rate impact, accelerated cyclic aging, and multiple fault coupling.

[0004] To address the aforementioned shortcomings, a technical solution is provided. Summary of the Invention

[0005] The purpose of this invention is to solve the problems of inaccurate fault diagnosis and difficulty in locating the root cause of faults in lithium batteries of communication base stations, and to propose a method, device and electronic equipment for fault diagnosis of lithium batteries of communication base stations.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] Methods for diagnosing lithium battery faults in communication base stations include:

[0008] S1. Multi-dimensional data acquisition: Through the battery management unit deployed on the lithium battery pack of the communication base station, the voltage, current, temperature and electrochemical impedance spectroscopy data of each individual cell in the battery pack are collected simultaneously.

[0009] S2. Inconsistency Feature Quantification: Based on multi-dimensional data, extract the static and dynamic inconsistency features of the battery pack, construct the comprehensive feature vector of each individual battery cell, and calculate the comprehensive inconsistency index.

[0010] S3. Anomaly Identification and Fault Matching: The comprehensive feature vector is matched with a preset fault feature matching library to identify the fault type and assess the severity level of the fault.

[0011] S4. Fault Root Cause Location: Based on the spatial distribution of the faulty individual, electrical topology correlation, impedance spectrum in-depth analysis and thermal behavior analysis, a comprehensive determination of the fault root cause mechanism is made.

[0012] As a further improvement of the present invention, the specific implementation process of the static inconsistency feature includes:

[0013] All data points where the battery pack is in a static state and the state of charge is in a stable range are selected; the average voltage of the battery pack is obtained by extracting the voltage of all individual cells at the same time stamp; and the voltage dispersion is obtained based on the voltage of individual cells and the average voltage of the battery pack.

[0014] The internal resistance of each individual cell at a preset frequency is measured using the AC impedance method. The internal resistance deviation coefficient is obtained based on the ratio of the internal resistance of each individual cell to the average internal resistance of the battery pack.

[0015] The state of charge (SOC) value of each individual cell is obtained based on the open-circuit voltage combined with the pre-calibrated open-circuit voltage SOC curve, and the range of the SOC values ​​of each individual cell is used as the SOC difference.

[0016] The internal resistance growth rate is obtained by comparing the initial internal resistance reference value of each individual cell with the current internal resistance measurement value.

[0017] The current actual usable capacity of each individual battery cell is obtained based on historical complete charge-discharge cycle data, and the capacity decay rate is obtained by comparing it with the rated capacity.

[0018] The comprehensive health status value is obtained by weighting and fusing the internal resistance growth rate and the capacity decay rate;

[0019] A static inconsistency vector for the battery pack is constructed based on voltage dispersion, internal resistance deviation coefficient, charge difference, and comprehensive health status value.

[0020] As a further improvement of the present invention, the specific implementation process of the dynamic inconsistency feature includes:

[0021] When the battery pack is in a dynamic operating state, extract dynamic response characteristics, response delay characteristics, and thermal imbalance characteristics.

[0022] The dynamic response characteristics are obtained based on the slope change of the voltage offset trajectory of each individual cell within the charge / discharge segment;

[0023] The response delay characteristics are obtained based on the delay time for the voltage of each individual cell to reach steady state when the load current changes abruptly.

[0024] Thermal imbalance feature extraction: Based on the time series data of surface temperature of each individual cell and the total current of the battery pack, the real-time temperature rise rate and heat generation characteristic coefficient are obtained, and the thermal imbalance feature is obtained based on the temperature rise rate and heat generation characteristic coefficient.

[0025] Based on dynamic response characteristics, response delay characteristics, and thermal imbalance characteristics, a dynamic inconsistency vector is constructed and merged with the static inconsistency vector to obtain a comprehensive feature vector.

[0026] As a further improvement of the present invention, the specific operation steps of the comprehensive inconsistency index are as follows:

[0027] The arithmetic mean of the comprehensive feature vectors of individual cells that have been confirmed to be in normal operation without faults for nearly 3 months after the initial commissioning of the battery pack is selected as the reference benchmark vector.

[0028] Based on the latest data, the comprehensive feature vector of each individual battery at the current moment is obtained, and the difference vector is obtained by the difference between the vector and the reference vector. Based on the same historical normal operation cycle, the historical comprehensive feature vector data of all individual batteries are extracted, and the overall covariance matrix at the battery pack level is constructed to obtain the feature deviation.

[0029] Extract the characteristic deviation of individual cells every hour during the review period to construct a deviation time series; set a sliding time window, assign weights to each historical data point within the window, and introduce a time decay factor to obtain the cumulative deviation trend value;

[0030] Based on the comprehensive feature vectors of all individual cells at the current evaluation time, a feature correlation coefficient matrix is ​​formed, and the local coupling anomaly degree is obtained based on the absolute difference between the local average correlation and the global average correlation.

[0031] Based on feature deviation, cumulative deviation trend value, and local coupling anomaly, a comprehensive inconsistency index is obtained by normalization and weighted fusion formula.

[0032] As a further improvement of the present invention, the specific operation steps of S3 are as follows:

[0033] Three characteristic frequency points with different internal state changes of the battery are selected, including the high frequency region, the medium frequency region and the low frequency region; the AC impedance value at the characteristic frequency point is extracted for each cell in the battery pack, and the average value and complex deviation of the impedance of all cells in the battery pack are obtained to form the cell impedance deviation spectrum.

[0034] Complex deviations are decomposed into real and imaginary deviation components. Preliminary fault location is then performed by combining the index change trends of static and dynamic inconsistency vectors, including abnormal connection impedance, internal aging differences, abnormal deterioration of solid electrolyte interface film, abnormal charge transfer obstruction, and complex aging anomalies.

[0035] A fault feature matching library is constructed, which includes various typical fault modes and their corresponding multi-dimensional feature templates. The comprehensive feature vector of the current single cell is matched with the fault feature templates in the fault feature matching library. The fault mode with the highest matching similarity is selected as the preliminary fault diagnosis result. The fault severity index is obtained based on the comprehensive inconsistency index, the fault type identification confidence, and the cumulative deviation trend value, and the fault severity level is generated.

[0036] As a further improvement of the present invention, the specific operation steps of S4 are as follows:

[0037] Based on the binding relationship between temperature sensor number and individual battery number, a mapping table between individual battery number and three-dimensional physical installation location is constructed. The identified faulty cells are mapped to a three-dimensional spatial coordinate system to form a distribution map. The spatial distribution characteristics and corresponding causes of the faulty cells are determined by the spatial clustering degree.

[0038] Based on the series and parallel electrical topology, the positional relationship of the faulty individual in the circuit is analyzed to obtain the correlation anomaly degree of the series branch and the current distribution deviation rate of the parallel branch, thereby identifying the risk of branch performance degradation and current imbalance.

[0039] The electrochemical impedance spectroscopy data of the faulty individual were fitted to the equivalent circuit network to obtain the values ​​of each impedance component and compared with the benchmark value. The root cause mechanism of the fault was determined by combining the impedance growth rate.

[0040] Construct a temperature, time and current, and time synchronization comparison chart to obtain the heat generation coefficient per unit current and the heat generation deviation rate, and determine the risk of heat-related faults.

[0041] A decision matrix is ​​constructed, and the support scores of each analysis dimension are weighted and integrated to obtain the comprehensive support score. The root cause of the final fault or the complex fault is determined, and the complete location record is stored in association.

[0042] A second aspect of the present invention provides a fault diagnosis device for lithium batteries in communication base stations, comprising:

[0043] Multi-dimensional data acquisition equipment: used to be deployed in the battery pack to simultaneously collect voltage, current, temperature and electrochemical impedance spectroscopy data of each individual cell, and communicate with the characteristic and state analysis equipment;

[0044] Feature and status analysis equipment: used to receive collected data, extract static and dynamic inconsistency features and calculate a comprehensive inconsistency index, match it with a fault feature database to identify fault type and level, and communicate with fault location equipment;

[0045] Fault location equipment: used to receive fault signals, construct a multi-dimensional judgment network through spatial distribution, electrical topology, impedance spectrum analysis and thermal behavior analysis to determine the core risk source and fault root cause, generate a visual traceability report, and communicate with the operation and maintenance decision equipment;

[0046] Operation and maintenance decision-making equipment: used to generate differentiated operation and maintenance work orders based on risk level, record handling data and feed it back to the management interface, and to optimize feature thresholds and model weights.

[0047] A third aspect of the present invention provides an electronic device for diagnosing lithium battery faults in communication base stations, comprising:

[0048] Central Processing Unit (CPU): The CPU is configured to execute computer programs stored in memory, thereby controlling and coordinating the workflow of various hardware modules.

[0049] Memory: Memory is coupled to the central processing unit and includes non-volatile memory and volatile memory;

[0050] Data acquisition interface: Serving as a bridge for communication with the field battery management unit, it typically includes an analog-to-digital converter and a communication controller;

[0051] Power module: Provides a stable and reliable power supply for the entire electronic device. It typically has a wide voltage input range and surge protection capabilities to adapt to the complex power supply environment of communication base stations.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] This invention achieves comprehensive and synchronous sensing of the electrical, thermal, and electrochemical characteristics of battery packs. It employs a comprehensive feature vector representation method that integrates static and dynamic inconsistencies, combining feature deviation based on Mahalanobis distance, time-decay-weighted cumulative deviation trend values, and local coupling anomalies based on spatial topology to construct a multi-level comprehensive inconsistency index. In the fault identification stage, it utilizes impedance spectral feature decomposition and a vector similarity matching mechanism from a fault feature matching library to rapidly locate fault types and assess confidence levels. In the fault root cause localization stage, it integrates diagnostic results from four dimensions: three-dimensional spatial clustering analysis, series-parallel electrical topology correlation analysis, equivalent circuit impedance component analysis, and thermal behavior anomaly analysis. A weighted decision matrix is ​​used to comprehensively determine the fault root cause mechanism and effectively identify complex faults. This significantly improves the comprehensiveness, accuracy, timeliness, and interpretability of fault root cause tracing for lithium battery packs in communication base stations. Attached Figure Description

[0054] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example:

[0057] like Figure 1 As shown, the fault diagnosis method for lithium batteries in communication base stations includes multi-dimensional data acquisition, inconsistency feature quantification, anomaly identification and fault matching, and fault root cause location.

[0058] S1. Multi-dimensional data collection:

[0059] The battery management unit, deployed on the lithium battery pack of the communication base station, serves as the core data acquisition unit, performing synchronous, high-frequency data acquisition tasks. The battery management unit includes a voltage acquisition circuit, a current sensor, a temperature sensor array, an AC impedance spectroscopy measurement module, and environmental monitoring sensors.

[0060] The voltage acquisition circuit synchronously measures the positive and negative voltages of each individual cell in the battery pack through a high-precision analog-to-digital converter.

[0061] Meanwhile, a Hall current sensor is used to monitor the total charging and discharging current of the battery pack in real time and record the direction and value of the current.

[0062] The temperature sensor array consists of multiple digital temperature sensors that are physically attached to the surface of the monitored battery cell to collect surface temperature data for each cell and record the binding relationship between the temperature sensor number and the corresponding battery cell number.

[0063] When the battery pack is in a static or stable float charging state, the AC impedance spectroscopy measurement module, controlled by the battery management unit, injects a series of small-amplitude AC current signals of specific frequencies into the target battery cells at regular intervals; simultaneously, it measures the voltage response of the target battery cells to the AC current signals; based on the amplitude ratio and phase difference between the voltage response and the AC current signal, it obtains the AC impedance of the target battery cells at each specific frequency point, forming a set of characteristic data reflecting the internal electrochemical state of the battery, namely, electrochemical impedance spectroscopy data; the specific frequencies refer to multiple frequency points selected according to a logarithmic law from 1kHz to 0.1Hz.

[0064] The system continuously collects environmental parameters of the battery pack through independent environmental monitoring sensors, including ambient temperature and relative humidity inside the installation cabinet, as well as alarm status signals provided by smoke and water immersion sensors.

[0065] All collected voltage, current, temperature, electrochemical impedance spectroscopy data, and environmental parameters, along with high-precision timestamps, are packaged together, stored in local non-volatile memory, and uploaded to the base station monitoring center in real time.

[0066] S2, Inconsistency Feature Quantification:

[0067] S201. Static Inconsistency Feature Extraction:

[0068] Based on the battery management unit, all data points in the stored raw data are selected where the battery pack is in a static state and the state of charge is in a stable range of 40%-60%; static state means that the absolute value of the total current is less than the preset threshold and lasts for more than 1 hour.

[0069] For all selected data points, extract the voltage of all individual cells at the same time stamp, and perform statistical analysis to obtain the average voltage of the battery pack at the current time stamp; then, using the formula... The voltage dispersion of a single cell is calculated; among which, Indicates the first Open-circuit voltage of a single cell This indicates the average voltage of the battery pack;

[0070] The internal resistance of each individual cell at a preset frequency is measured using the AC impedance method. Based on the ratio of the internal resistance of each individual cell to the average internal resistance of the battery pack, the internal resistance deviation coefficient is obtained.

[0071] Based on the open-circuit voltage combined with the pre-calibrated open-circuit voltage state-of-charge curve, the state-of-charge value of each individual cell is obtained, and the range of the state-of-charge values ​​of each individual cell is used as the degree of charge difference.

[0072] Obtain the initial internal resistance reference value and the current internal resistance measurement value of each individual cell, and compare them to obtain the internal resistance growth rate;

[0073] Based on the complete historical charge and discharge cycle data recorded by the battery management unit, the current actual usable capacity of each individual battery cell is obtained by combining the operating condition analysis method, and the capacity decay rate is obtained by comparing it with the rated capacity of the battery.

[0074] The overall health status value of a single cell is obtained by weighting and fusing the internal resistance growth rate and the capacity decay rate.

[0075] Based on voltage dispersion, internal resistance deviation coefficient, charge difference, and comprehensive health status value, a static inconsistency vector of the battery pack is constructed.

[0076] S202, Dynamic Inconsistency Feature Extraction:

[0077] When the battery pack is detected to be in a dynamic charging / discharging state, dynamic feature extraction is performed:

[0078] Dynamic response feature extraction: continuously collect the voltage and total current of each individual battery cell within a complete charge-discharge segment at a preset sampling frequency; obtain the voltage offset trajectory of each individual battery cell based on the average voltage of the battery pack;

[0079] Based on the characteristic change points of the charge / discharge rate as the boundary, the voltage offset trajectory is automatically segmented, and the slope of the voltage offset trajectory of the corresponding segment is obtained by linear fitting of the data in each segment; the characteristic change points include changes in current direction and inflection points of step rise or fall.

[0080] After completing the piecewise linear fitting of the voltage offset trajectory, for the first... For each individual cell, within a complete charge-discharge cycle, a sequence label consisting of the fitted slopes of each segment will be obtained. ;

[0081] Through formula The dynamic response characteristic values ​​are obtained through calculation; where, This represents the total number of segments within a complete charge / discharge cycle. Indicates the first The first single cell Segmented voltage offset rate, Represents a sequence The arithmetic mean;

[0082] The larger the dynamic response characteristic value, the more unstable the voltage offset rate of the corresponding single cell is with load changes, that is, the worse the consistency of the dynamic response.

[0083] Response delay feature extraction: Capturing the moment when the load current undergoes a step change based on the battery management unit. Record the moment when the current step occurs. The instantaneous value of the corresponding single-cell terminal voltage is used as the initial voltage;

[0084] From the moment when the step change occurs Initially, voltage time-series data is continuously recorded. Starting from time t, a preset steady-state determination time window is set, and the voltage data points within the time window are linearly fitted to obtain the average voltage change rate. The voltage fluctuation amplitude is obtained based on the difference between the maximum and minimum voltage values ​​within the time window.

[0085] If both the average voltage change rate and the voltage fluctuation amplitude are greater than the corresponding preset thresholds, the voltage is determined to have entered a steady state, and the average voltage value during the current period is marked as the new steady-state value. The total voltage change of this step response is obtained based on the difference between the steady-state value and the initial voltage. A preset percentage point of the total change is used as the target change; for example, the target change is set to 90% of the total change.

[0086] From the moment when the step change occurs Initially, when the voltage change rate first reaches or exceeds the target change amount, the current time is recorded as... ;

[0087] Based on time With time The difference is used to obtain the absolute response delay time; the average delay time of the battery pack is obtained by averaging the absolute response delay times of all individual cells.

[0088] The standardized response delay characteristics are obtained by comparing the absolute response delay time of a single cell with the average delay time of the battery pack.

[0089] Thermal Imbalance Feature Extraction:

[0090] Based on the time-series data of surface temperature of each individual cell and the total current of the battery pack; after smoothing the time-series data of surface temperature of each individual cell, the real-time temperature rise rate is obtained based on the difference of temperature values ​​between the sampling time intervals; and the heat generation characteristic coefficient is obtained based on the temperature rise rate under unit current.

[0091] Select a complete charge-discharge segment, the same as the one used in the dynamic voltage response analysis, and obtain the average heat generation coefficient of each cell based on the time average of the heat generation characteristic coefficient of each cell within the current segment.

[0092] The average heat generation characteristic coefficient of the battery pack is obtained by averaging the average heat generation coefficients of all individual cells in the current segment.

[0093] The thermal imbalance characteristics are obtained based on the relative deviation ratio of the average heat generation coefficient of each individual cell relative to the average level of the battery pack.

[0094] Based on dynamic response characteristics, response delay characteristics, and thermal imbalance characteristics, a dynamic inconsistency vector is constructed and merged with the static inconsistency vector to form the... The comprehensive feature vector of each individual cell.

[0095] S203, Calculation of the Comprehensive Inconsistency Index:

[0096] Feature deviation: Select historical data of the battery pack in the past 3 months after the initial commissioning and confirmed normal operation cycle without faults. Perform S201 and S202 on each individual cell in the current normal operation cycle to obtain the corresponding comprehensive feature vector. Calculate the arithmetic mean of the comprehensive feature vectors of all individual cells obtained in the normal operation cycle, and use the calculation result as the reference benchmark vector of the corresponding individual cell.

[0097] Based on the latest data collected by S1 and the feature extraction processes of S201 and S202, the comprehensive feature vector of each individual battery at the current moment is obtained.

[0098] The difference vector is obtained based on the difference between the comprehensive feature vector at the current moment and the reference vector;

[0099] Based on the same historical normal operation cycle used to establish the reference benchmark vector, historical comprehensive feature vector data of all individual cells are extracted to construct the overall covariance matrix at the battery pack level.

[0100] Through formula The feature deviation is calculated, where, Represents the difference vector. This represents the transpose of the difference vector. The matrix representing the inverse of the population covariance matrix;

[0101] Cumulative Deviation Trend Value: A pre-set review period is used to extract the first... The deviation time series is constructed by analyzing the characteristic deviation of each individual cell per hour within the review period.

[0102] A sliding time window is defined, and a weight is assigned to each historical data point within the sliding time window. A time decay factor is introduced, and the formula is used to calculate the weight. The cumulative deviation trend value is calculated; where, Indicates the length of the sliding time window. Indicates the first The single cell is in the first The characteristic deviation of the step, Indicates the first The influence weighting factor of the feature deviation corresponding to the step;

[0103] Coupling anomaly: Based on the comprehensive feature vector of all individual cells at the current evaluation time, calculate the Pearson correlation coefficient between any two individual cell comprehensive feature vectors to form a feature correlation coefficient matrix at the battery pack level;

[0104] For the For each individual cell, based on the physical layout and electrical connection topology of the battery pack, the set of all corresponding adjacent cells is determined, including cells that are directly adjacent in physical space and cells that are directly connected in series or parallel in electrical space; the average Pearson correlation coefficient between the comprehensive feature vector of all individual cells at the current evaluation time and the comprehensive feature vector of all neighboring individual cells in the set of adjacent cells is used as the local average correlation.

[0105] The global average correlation of the battery pack is obtained by taking the arithmetic mean of the local average correlations of all elements in the feature correlation coefficient matrix.

[0106] Based on the The absolute difference between the local average correlation and the global average correlation of a single cell is used to obtain the local coupling anomaly.

[0107] Based on feature deviation, cumulative deviation trend value, and local coupling anomaly, a comprehensive inconsistency index is obtained by normalization and weighted fusion formula.

[0108] S3. Anomaly identification and fault matching:

[0109] Based on the multi-dimensional data collected by the battery management unit from S1, three characteristic frequency points with different changes in the internal state of the battery are selected, including high, medium and low frequency regions; for example, the high frequency region is 1kHz, which mainly reflects ohmic impedance, the medium frequency region is 10Hz, which mainly reflects the impedance of the solid electrolyte interface film, and the low frequency region is 0.1Hz, which mainly reflects the charge transfer impedance.

[0110] For each individual cell in the battery pack, the AC impedance value of each individual cell at the characteristic frequency point is extracted and recorded in complex form, including the real part and the imaginary part.

[0111] For each characteristic frequency point, the formula is used. The average impedance of all individual cells in the battery pack is calculated, where, Indicates the first Each single entity at a characteristic frequency The real part of the impedance below, Indicates the first Each single entity at a characteristic frequency The imaginary part of the impedance below, This indicates the total number of individual battery cells. The imaginary unit is used to distinguish between the real and imaginary parts of a complex number.

[0112] Through formula The complex deviation is calculated. These represent the arithmetic mean of the real and imaginary parts of the impedance of all individual units, respectively.

[0113] Based on the complex deviation of each individual cell at all characteristic frequency points, a single cell impedance deviation spectrum is formed;

[0114] The complex deviation is decomposed into a real deviation component reflecting the difference in ohmic internal resistance and an imaginary deviation component reflecting the difference in electrochemical polarization.

[0115] If the real deviation of a single cell exceeds the preset threshold range and the voltage dispersion index in the static inconsistency vector shows a continuous upward trend, it is initially identified as an abnormal connection impedance.

[0116] If the imaginary part deviation of a single cell exceeds the preset threshold range and the dynamic inconsistency vector is lower than the preset threshold, it is initially identified as an internal aging difference.

[0117] If the imaginary part deviation of a single cell in the mid-frequency region shows a continuous increasing trend, and the internal resistance growth rate in the static inconsistency vector exceeds a preset threshold, it is initially identified as an abnormal degradation of the solid electrolyte interface film.

[0118] If the impedance modulus of a single cell in the low-frequency region deviates from the average level of the battery pack, and the thermal imbalance characteristics in the dynamic inconsistency vector exceed a preset threshold, it is initially identified as an abnormality of charge transfer obstruction.

[0119] If both the real and imaginary deviations of a single cell exceed the preset threshold range, and the cumulative deviation trend value in the comprehensive inconsistency index shows an accelerating upward trend, then it is initially identified as a composite aging anomaly.

[0120] A fault feature matching library is constructed, which contains a variety of typical fault modes and corresponding multi-dimensional feature templates. Typical fault modes include internal micro short circuit risk, thermal runaway precursor, capacity drop, loose tab connection, electrolyte drying and lithium dendrite growth.

[0121] For each fault mode, the typical feature combinations of the fault mode on the static inconsistency vector, dynamic inconsistency vector, and individual impedance deviation spectrum are pre-calibrated to form a fault feature template; the fault feature template is represented in vector form, and the dimension is consistent with the comprehensive feature vector.

[0122] The current single cell's comprehensive feature vector and single cell impedance deviation spectrum are matched with each fault feature template in the fault feature matching library based on similarity.

[0123] Through formula Calculations are performed to obtain the current single cell and the first... The similarity of matching fault modes; where, Indicates the first The current comprehensive feature vector of each individual cell, Indicates the first Fault feature template vectors for various fault modes;

[0124] The fault mode with the highest matching similarity is selected as the preliminary fault diagnosis result of the current single cell.

[0125] The maximum value of the matching similarity is used as the confidence level for fault type identification. If the maximum value of the matching similarity is less than the preset minimum confidence threshold, the current single cell is marked as an unknown fault type, and the corresponding comprehensive feature vector is stored in the queue to be manually reviewed.

[0126] After normalization based on the comprehensive inconsistency index, fault type identification confidence, and cumulative deviation trend value, the fault severity index is obtained through a weighted fusion formula.

[0127] A fault severity level is generated based on a fault severity index, including four levels: Normal, Attention, Warning, and Emergency.

[0128] The fault type, fault level, fault type identification confidence level, and fault severity index are associated and stored to form a fault identification result record.

[0129] S4. Fault Root Cause Location:

[0130] S401, Faulty Individual Unit Spatial Location:

[0131] Based on the binding relationship between the temperature sensor number and the individual battery number recorded in the battery management unit in S1, a mapping table between the individual battery number and the physical installation location is constructed; the physical installation location is described using a three-dimensional coordinate system, including the layer number, column number, and serial number position information of the battery rack in a single column;

[0132] All faulty individual cells identified in S3 are mapped to the three-dimensional spatial coordinate system of the battery pack to form a spatial distribution map of faulty individual cells;

[0133] Through formula Spatial clustering is calculated; where, Indicates the number of faulty units. Indicates the relationship between the i-th single cell and the i-th cell. The three-dimensional Euclidean distance between individual cells Indicates the standard spacing between adjacent cells within the battery pack;

[0134] When the spatial clustering degree is less than the preset threshold range, it is determined that the faulty single cell exhibits spatial clustering characteristics, indicating that there is a regional fault caused by local environmental factors, including poor local heat dissipation, excessively high local temperature, or abnormal local humidity.

[0135] When the spatial clustering is within a preset threshold range, the faulty individual cells are judged to exhibit random distribution characteristics, indicating random faults caused by individual differences in individual cells or differences in manufacturing batches.

[0136] When the spatial clustering exceeds the preset threshold range, the faulty individual cells are determined to exhibit a dispersed distribution characteristic, indicating a general problem caused by abnormal overall operating conditions.

[0137] S402, Electrical Topology Correlation Analysis:

[0138] Based on the series and parallel electrical topology diagram of the battery pack, the positional relationship and mutual influence of the faulty cells in the electrical circuit are analyzed.

[0139] For individual cells connected in series, identify the series branch number where the faulty individual cell is located; check the overall inconsistency index of other individual cells in the same series branch;

[0140] The correlation anomaly degree of the series branch is obtained based on the comprehensive inconsistency index of the individual cells in the same series branch and the arithmetic mean of the comprehensive inconsistency index of all individual cells in the battery pack.

[0141] When the correlation anomaly of the series branch exceeds the preset threshold, it is determined that there is a risk of overall performance degradation in the current series branch; other individual cells in the current branch, except for the identified faulty cells, are marked as key monitoring targets.

[0142] For modules connected in parallel, analyze the current distribution balance of each branch in the parallel module where the faulty unit is located.

[0143] Based on the equivalent internal resistance of each parallel branch, the current distribution within the parallel module is obtained; using the formula... Calculation yields the first The current distribution ratio of each parallel branch; among which... Indicates the first The equivalent internal resistance of a parallel branch, Indicates the total number of parallel branches. Indicates the first The equivalent internal resistance of each parallel branch;

[0144] Through formula Calculation yields the first Current distribution deviation rate of parallel branches;

[0145] When the current distribution deviation rate exceeds the preset threshold, the corresponding branch is marked as having a risk of current imbalance; the current imbalance risk factors are included in the fault root cause analysis record.

[0146] S403, In-depth analysis of impedance spectrum: Based on the complete electrochemical impedance spectroscopy data collected by the AC impedance spectroscopy measurement module in S1, in-depth analysis of impedance spectrum is performed on the faulty single cell.

[0147] The electrochemical impedance spectroscopy data of the faulty single cell are fitted to an equivalent circuit network; the equivalent circuit network includes a series ohmic resistor, a solid electrolyte interface membrane impedance unit, and a charge transfer impedance unit.

[0148] The parameters of the equivalent circuit network are fitted and optimized using the nonlinear least squares method to obtain the values ​​of each impedance component, including ohmic impedance, solid electrolyte interface membrane impedance, and charge transfer impedance.

[0149] Obtain the reference values ​​of each impedance component of the faulty individual cell calibrated at the initial commissioning;

[0150] The growth rates of ohmic impedance, solid electrolyte interfacial membrane impedance, and charge transfer impedance are obtained by comparing them with the corresponding impedance component reference values.

[0151] Construct a rule base for the correspondence between impedance component variation characteristics and fault root cause mechanisms:

[0152] When the growth rate of ohmic impedance is greater than the growth rate of solid electrolyte interface film impedance and the growth rate of charge transfer impedance, and the growth rate of ohmic impedance exceeds the preset threshold, the root cause of the fault is determined to be the current collector corrosion, the deterioration of the tab welding point, or the blockage of the ion conduction path caused by the drying of the electrolyte.

[0153] When the growth rate of the solid electrolyte interface film impedance is greater than the growth rate of the ohmic impedance and the growth rate of the charge transfer impedance, and the growth rate of the solid electrolyte interface film impedance exceeds the preset threshold, the root cause of the fault is determined to be abnormal thickening of the solid electrolyte interface film, the early stage of lithium dendrite growth, or the aggravation of side reactions on the negative electrode surface.

[0154] When the charge transfer impedance growth rate is greater than the ohmic impedance growth rate and the solid electrolyte interface film impedance growth rate, and the charge transfer impedance growth rate exceeds the preset threshold, the root cause of the fault is determined to be the deterioration of the positive electrode active material structure, the phase change of the electrode material, or the obstruction of lithium ion insertion and extraction kinetics.

[0155] When all impedance growth rates increase and the difference between them is less than a preset threshold, the root cause of the fault is determined to be overall aging or cycle life depletion.

[0156] S404, Analysis of Abnormal Thermal Behavior:

[0157] Construct a synchronous comparison diagram of temperature-time curves and current-time curves for faulty individual cells;

[0158] The heat generation coefficient per unit current is obtained by the ratio of the temperature rise rate to the current of a faulty single cell under typical charge and discharge conditions.

[0159] The heat generation deviation rate is obtained by comparing the heat generation coefficient per unit current of the faulty individual cell with the average heat generation coefficient of the battery pack.

[0160] When the heat generation deviation rate continues to exceed the preset threshold range and the cumulative deviation trend value shows an accelerating upward trend, it is determined that there is an internal micro short circuit risk or a precursor to thermal runaway.

[0161] When the heat generation deviation rate is within the preset threshold range and the real part of the impedance deviation is higher than the preset threshold, it is determined that there is an aggravation of the Ohmic thermal effect caused by an abnormal increase in internal resistance.

[0162] When the heat generation deviation rate is less than the preset threshold range, it is determined that the thermal behavior is within the normal fluctuation range and the thermal factor is not the main source of failure.

[0163] S405, Comprehensive Determination of Fault Root Causes:

[0164] A fault root cause determination decision matrix is ​​constructed. The rows of the decision matrix correspond to six typical fault root cause mechanisms, namely current collector corrosion, solid electrolyte interface film thickening, active material degradation, internal micro-short circuit, connection point abnormality, and overall aging. The columns of the decision matrix correspond to the analysis results of five analysis dimensions, including fault type identification, fault individual spatial location, electrical topology correlation analysis, impedance spectrum in-depth analysis, and thermal behavior anomaly analysis.

[0165] For each analytical dimension, the degree of support for various root cause mechanisms of failures is scored based on the analytical results; this is achieved through a formula. Calculation yields the first The overall support for the root cause mechanism of the type of failure; among which, Indicates the total number of analysis dimensions. Indicates the first Weight coefficients for each analytical dimension, Indicates the first The analytical dimension for the first Support score for the root cause mechanism of the type of failure;

[0166] The fault root cause mechanism with the highest overall support was selected as the final judgment result; the overall support value was used as the fault root cause localization confidence.

[0167] When the difference in the overall support of two or more fault root cause mechanisms is less than a preset threshold, it is judged as a composite fault; the two fault root cause mechanisms with the highest overall support are arranged in descending order, recorded together in the diagnostic conclusion, and marked as multi-factor coupled fault.

[0168] The faulty unit number, physical installation location, fault type, fault level, fault root cause mechanism, and fault root cause location reliability are associated and stored to form a complete fault root cause location record.

[0169] This invention is a fault diagnosis device for lithium batteries in communication base stations, comprising:

[0170] Multi-dimensional data acquisition equipment: synchronously measures the terminal voltage of each individual battery cell through a voltage acquisition circuit; monitors the total current of the battery pack using a Hall current sensor; collects surface temperature data through a temperature sensor array attached to the battery surface and binds the sensors to the battery number; injects a specific frequency AC signal into the battery in a static or float charging state to measure and calculate the impedance spectrum data of each individual cell; all data are accompanied by a high-precision timestamp, stored locally and then uploaded.

[0171] Feature and State Analysis Equipment: At the static level, a static inconsistency vector is constructed based on voltage dispersion, internal resistance deviation coefficient, charge difference, and overall health status. At the dynamic level, the slope of the voltage offset trajectory, the current step response delay time, and thermal imbalance features during the charging and discharging process are extracted to form a dynamic inconsistency vector. A comprehensive inconsistency index is obtained based on the current feature deviation degree based on Mahalanobis distance, the historical cumulative deviation trend based on time decay weighting, and the local coupling anomaly degree reflecting topological association. The comprehensive feature vector of a single cell is compared with a pre-built fault feature matching library to identify the fault type. The fault severity level is generated based on the comprehensive inconsistency index and confidence level.

[0172] Fault location equipment: Through spatial location, topological correlation analysis, impedance spectrum in-depth analysis, and thermal behavior analysis, parallel analysis is performed, the analysis results are summarized for comprehensive judgment, and a decision matrix is ​​constructed for weighted scoring to determine the core fault root cause;

[0173] Operation and maintenance decision-making equipment: Based on the risk level determined by the positioning equipment, it automatically generates differentiated operation and maintenance strategies, dispatches operation and maintenance work orders, and records the operation process and handling effect of operation and maintenance personnel throughout the process, dynamically calibrates the baseline threshold for fault judgment and optimizes feature weight allocation.

[0174] This invention relates to an electronic device for diagnosing lithium battery faults in communication base stations, comprising:

[0175] Central Processing Unit (CPU): The CPU is configured to execute computer programs stored in memory, thereby controlling and coordinating the workflow of various hardware modules; algorithms for running fault diagnosis methods, including calculating static and dynamic inconsistency feature vectors and comprehensive inconsistency index, performing similarity matching of fault feature matching library, coordinating multi-dimensional analysis of fault root cause localization, and performing comprehensive fault root cause determination.

[0176] Memory: The memory is coupled to the central processing unit and includes non-volatile memory and volatile memory; the non-volatile memory is used to store multi-source data, including computer program instructions for fault diagnosis methods, preset fault feature matching library, equivalent circuit network parameters, historical and real-time acquired multi-dimensional battery data, extracted feature vectors, diagnostic result records and fault root cause location records;

[0177] Data acquisition interface: Serving as a bridge for communication with the field battery management unit, it typically includes an analog-to-digital converter and a communication controller; it is configured to: receive raw data from multiple sensors deployed on the battery pack; and receive electrochemical impedance spectroscopy data from the AC impedance spectroscopy measurement module to acquire the electrochemical impedance spectroscopy data of each individual cell.

[0178] Power module: Used to provide a stable and reliable power supply for the entire electronic device. It usually has a wide voltage input range and surge protection capability to adapt to the complex power supply environment of communication base stations.

[0179] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for diagnosing lithium battery faults in communication base stations, characterized in that, include: S1. Multi-dimensional data acquisition: Through the battery management unit deployed on the lithium battery pack of the communication base station, the voltage, current, temperature and electrochemical impedance spectroscopy data of each individual cell in the battery pack are collected simultaneously. S2. Inconsistency Feature Quantification: Based on multi-dimensional data, extract the static and dynamic inconsistency features of each individual cell, construct a comprehensive feature vector for each individual cell, and calculate the comprehensive inconsistency index; among them, the static inconsistency vector of each individual cell is constructed based on voltage dispersion, internal resistance deviation coefficient, charge difference, and comprehensive health state value; the dynamic inconsistency vector of each individual cell is constructed based on dynamic response features, response delay features, and thermal imbalance features. The specific operational steps for implementing the comprehensive inconsistency index are as follows: The arithmetic mean of the comprehensive characteristic vectors of each individual cell in the historical normal operation cycle confirmed to be fault-free for nearly 3 months after the initial commissioning of the battery pack was completed was selected as the reference benchmark vector. Based on the latest data collected by S1, the comprehensive feature vector of each individual battery at the current moment is obtained. The difference vector is obtained based on the difference between the comprehensive feature vector at the current moment and the reference vector. Based on the same historical normal operation cycle, the historical comprehensive feature vector data of all individual batteries are extracted, and the overall covariance matrix at the battery pack level is constructed to obtain the feature deviation. A pre-defined review period is used to extract the hourly characteristic deviation of individual cells within the review period to construct a deviation time series; a sliding time window is set, and weights are assigned to each historical data point within the window, and a time decay factor is introduced to obtain the cumulative deviation trend value. Based on the comprehensive feature vectors of all individual cells at the current moment, a feature correlation coefficient matrix is ​​formed, and the local coupling anomaly degree is obtained based on the absolute difference between the local average correlation and the global average correlation. Based on feature deviation, cumulative deviation trend value and local coupling anomaly, a comprehensive inconsistency index is obtained by normalization and weighted fusion formula. S3. Anomaly Identification and Fault Matching: The comprehensive feature vector is matched with a preset fault feature matching library to identify the fault type and assess the severity level of the fault. The specific operation steps of S3 are as follows: Three characteristic frequency points with different internal state changes of the battery are selected, including the high frequency region, the medium frequency region and the low frequency region; the AC impedance value at the characteristic frequency point is extracted for each cell in the battery pack, and the average value and complex deviation of the impedance of all cells in the battery pack are obtained to form the cell impedance deviation spectrum. The complex deviation is decomposed into real and imaginary deviation components, and the index change trends of static and dynamic inconsistency vectors are combined to perform preliminary fault location. S4. Fault Root Cause Location: Based on the spatial distribution of the faulty individual, electrical topology correlation, impedance spectrum in-depth analysis and thermal behavior analysis, a comprehensive determination of the fault root cause mechanism is made.

2. The method for diagnosing lithium battery faults in communication base stations according to claim 1, characterized in that, The specific implementation process of the static inconsistency feature includes: All data points where the battery pack is in a static state and the state of charge is in a stable range are selected; the average voltage of the battery pack is obtained by extracting the voltage of all individual cells at the same time stamp; and the voltage dispersion is obtained based on the voltage of individual cells and the average voltage of the battery pack. The internal resistance of each individual cell at a preset frequency is measured using the AC impedance method. The internal resistance deviation coefficient is obtained based on the ratio of the internal resistance of each individual cell to the average internal resistance of the battery pack. The state of charge (SOC) value of each individual cell is obtained based on the open-circuit voltage combined with the pre-calibrated open-circuit voltage SOC curve, and the range of the SOC values ​​of each individual cell is used as the SOC difference. The internal resistance growth rate is obtained by comparing the initial internal resistance reference value of each individual cell with the current internal resistance measurement value. The current actual usable capacity of each individual battery cell is obtained based on historical complete charge-discharge cycle data, and the capacity decay rate is obtained by comparing it with the rated capacity. The comprehensive health status value is obtained by weighting and fusing the internal resistance growth rate and the capacity decay rate; Based on voltage dispersion, internal resistance deviation coefficient, charge difference, and comprehensive health status value, a static inconsistency vector for each individual cell is constructed.

3. The method for diagnosing lithium battery faults in communication base stations according to claim 1, characterized in that, The specific implementation process of the dynamic inconsistency feature includes: When the battery pack is in a dynamic operating state, extract dynamic response characteristics, response delay characteristics, and thermal imbalance characteristics. The dynamic response characteristics are obtained based on the slope change of the voltage offset trajectory of each individual cell within the charge / discharge segment; The response delay characteristics are obtained based on the delay time for the voltage of each individual cell to reach steady state when the load current changes abruptly. Thermal imbalance feature extraction: Based on the time series data of surface temperature of each individual cell and the total current of the battery pack, the real-time temperature rise rate and heat generation characteristic coefficient are obtained, and the thermal imbalance feature is obtained based on the temperature rise rate and heat generation characteristic coefficient. Based on dynamic response characteristics, response delay characteristics, and thermal imbalance characteristics, a dynamic inconsistency vector for each individual cell is constructed and merged with the static inconsistency vector to obtain a comprehensive feature vector.

4. The method for diagnosing lithium battery faults in communication base stations according to claim 1, characterized in that, The specific steps of S3 are as follows: Preliminary fault location includes: Abnormal connection impedance, internal aging differences, abnormal deterioration of solid electrolyte interface film, abnormal charge transfer obstruction, and complex aging anomalies. A fault feature matching library is constructed, which includes various typical fault modes and their corresponding multi-dimensional feature templates. The comprehensive feature vector of a single cell at the current moment is matched with each fault feature template in the fault feature matching library. The fault mode with the highest matching similarity is selected as the preliminary fault diagnosis result. The fault severity index is obtained based on the comprehensive inconsistency index, the fault type identification confidence, and the cumulative deviation trend value, and the fault severity level is generated.

5. The method for diagnosing lithium battery faults in communication base stations according to claim 1, characterized in that, The specific operation steps of S4 are as follows: Based on the binding relationship between temperature sensor number and individual battery number, a mapping table between individual battery number and three-dimensional physical installation location is constructed. The identified faulty cells are mapped to a three-dimensional spatial coordinate system to form a distribution map. The spatial distribution characteristics and corresponding causes of the faulty cells are determined by the spatial clustering degree. Based on the series and parallel electrical topology, the positional relationship of the faulty individual in the circuit is analyzed to obtain the correlation anomaly degree of the series branch and the current distribution deviation rate of the parallel branch, thereby identifying the risk of branch performance degradation and current imbalance. The electrochemical impedance spectroscopy data of the faulty individual were fitted to the equivalent circuit network to obtain the values ​​of each impedance component and compared with the benchmark value. The root cause mechanism of the fault was determined by combining the impedance growth rate. Construct a synchronous comparison chart of temperature-time curves and current-time curves for faulty individual cells, obtain the heat generation coefficient per unit current and the heat generation deviation rate, and determine the risk of heat-related faults. A decision matrix is ​​constructed, and the support scores of each analysis dimension are weighted and integrated to obtain the comprehensive support score. The root cause of the final fault or the complex fault is determined, and the complete location record is stored in association.

6. An apparatus for use in the fault diagnosis method for lithium batteries in communication base stations according to any one of claims 1-5, comprising: Multi-dimensional data acquisition equipment: used to be deployed in the battery pack to simultaneously collect voltage, current, temperature and electrochemical impedance spectroscopy data of each individual cell, and communicate with the characteristic and state analysis equipment; Feature and status analysis equipment: used to receive collected data, extract static and dynamic inconsistency features and calculate a comprehensive inconsistency index, match it with a fault feature database to identify fault type and level, and communicate with fault location equipment; Fault location equipment: used to receive fault signals, construct a multi-dimensional judgment network through spatial distribution, electrical topology, impedance spectrum analysis and thermal behavior analysis to determine the core risk source and fault root cause, generate a visual traceability report, and communicate with the operation and maintenance decision equipment; Operation and maintenance decision-making equipment: used to generate differentiated operation and maintenance work orders based on risk level, record handling data and feed it back to the management interface, and to optimize feature thresholds and model weights.

7. An electronic device applied to the fault diagnosis method for lithium batteries in communication base stations according to any one of claims 1-5, comprising: Central Processing Unit (CPU): The CPU is configured to execute computer programs stored in memory, thereby controlling and coordinating the workflow of various hardware modules. Memory: Memory is coupled to the central processing unit and includes non-volatile memory and volatile memory; Data acquisition interface: serving as a bridge for communication with the field battery management unit, including an analog-to-digital converter and a communication controller; Power module: Provides a stable and reliable power supply for the entire electronic device, with a wide voltage input range and surge protection capability to adapt to the complex power supply environment of communication base stations.

Images