Power distribution terminal backup power supply monitoring method and system based on insulation internal resistance online detection

Abstract

The invention discloses a power distribution terminal backup power supply monitoring method and system based on insulation internal resistance online detection, and relates to the technical field of power distribution terminal backup power supply monitoring. The voltage signal and the current signal comprise natural perturbation components generated by load change or start and stop of the communication module; dynamic response characteristics under perturbation excitation are extracted based on the collected voltage signals and current signals, temperature drift correction is carried out on the dynamic response characteristics in combination with the battery temperature measured in real time, and a corrected equivalent insulation internal resistance value is obtained; and establishing a battery connection topological graph model according to a battery series-parallel topological structure of a backup power supply in the power distribution terminal, inputting the corrected equivalent insulation internal resistance value and the dynamic response characteristic into a sparse system identification model, and decoupling the individual insulation internal resistance of each single battery.

Description

Technical Field

[0001] This invention relates to the field of backup power monitoring technology for power distribution terminals, and in particular to a method and system for monitoring backup power supply for power distribution terminals based on online detection of insulation resistance. Background Technology

[0002] Distribution terminals (such as FTUs, DTUs, and TTUs) are key field devices in distribution network automation systems, used to achieve functions such as telemetry, remote signaling, and remote control. To ensure that communication and control operations can be maintained even when the main power supply fails, distribution terminals are generally equipped with backup power supplies (usually lead-acid batteries or lithium iron phosphate battery packs). Distribution terminal backup power supply monitoring technology refers to the technical means of real-time or periodic assessment of the health status of these backup power supplies (especially insulation performance, capacity retention capability, and internal resistance changes).

[0003] For example, patent publication number CN119959787A discloses a method for managing the performance of backup power supply for a power distribution terminal. This method includes configuring a dehumidifier to regulate ambient humidity based on the surrounding environment of the backup power supply, and configuring a temperature control enclosure and / or attaching cooling materials to regulate ambient temperature. Based on the battery in the backup power supply, a power management module is configured on a designated monitoring host to monitor the battery voltage, internal resistance, and temperature online, as well as the ambient humidity. Furthermore, parameters for battery float charging, equalization charging, and activation are set and monitored online. Implementing this invention not only manages battery float charging, equalization charging, and activation, but also enables online monitoring of battery voltage, internal resistance, temperature, ambient humidity, and ambient temperature.

[0004] However, in existing technologies, backup power supplies for distribution terminals are usually composed of multiple batteries connected in series and parallel. Traditional methods only measure the parameters of the entire group of ports. When the insulation of a single cell deteriorates or its internal resistance becomes abnormal, the total voltage may still be within the normal range, resulting in strong fault concealment and difficulty in locating the fault. Furthermore, the internal resistance of batteries is highly sensitive to temperature, and existing online methods do not perform effective temperature compensation. They often misjudge the increase in internal resistance caused by low temperature as deterioration, resulting in a large number of false alarms and reducing the trust in operation and maintenance. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides a monitoring method for backup power supply of distribution terminals based on online insulation resistance detection. This addresses the problem that backup power supplies of distribution terminals are usually composed of multiple batteries connected in series and parallel. Traditional methods only measure the parameters of the entire group of ports. When the insulation of a single cell deteriorates or the internal resistance is abnormal, the total voltage may still be within the normal range, resulting in strong fault concealment and difficulty in fault location. Furthermore, battery internal resistance is highly sensitive to temperature, and existing online methods do not perform effective temperature compensation, often misjudging the increase in internal resistance caused by low temperature as deterioration, resulting in a large number of false alarms and reducing the trust of operation and maintenance personnel.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for monitoring backup power supply in a power distribution terminal based on online insulation resistance detection, comprising: The voltage and current signals of the backup power port of the power distribution terminal are collected under normal operating conditions. The voltage and current signals include natural perturbation components generated by load changes or the start and stop of the communication module. Based on the collected voltage and current signals, the dynamic response characteristics under perturbation excitation are extracted, and the dynamic response characteristics are corrected for temperature drift by combining the real-time measured battery temperature to obtain the corrected equivalent insulation internal resistance value. A battery connection topology model is established based on the battery series-parallel topology of the backup power supply in the power distribution terminal. The corrected equivalent insulation resistance and dynamic response characteristics are input into the sparse system identification model to decouple the individual insulation resistance of each cell. The individual insulation resistance of each cell is continuously sampled to form an individual insulation resistance time series. The information entropy and coefficient of variation of the time series are calculated within a sliding time window of a preset length. The internal resistance stability index is generated based on the information entropy and coefficient of variation. Obtain communication anomaly event records of the power distribution terminal within the time period synchronized with the sliding time window, and generate multidimensional health assessment indicators by combining the internal resistance stability index, the current value of individual insulation internal resistance, and the communication anomaly event records. When the internal insulation resistance of any single cell exceeds the adaptive dynamic threshold, or when the multidimensional health assessment indicators meet the preset warning criteria, an alarm message containing the location identifier of the corresponding single cell and maintenance suggestions is generated and output through the local interface or remote communication channel.

[0008] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the specific steps for collecting the voltage and current signals of the backup power port of the power distribution terminal under normal operating conditions are as follows: The high-speed data acquisition module inside the power distribution terminal is activated to synchronously acquire the terminal voltage signal and loop current signal between the positive and negative poles of the backup power supply at a fixed sampling frequency without interrupting the terminal's business functions. The acquired voltage and current signals are timestamped to form a synchronous timing data stream; Scan the synchronous timing data stream to identify current step change events caused by periodic wake-up of the communication module, telemetry status flipping, or external load access; For each identified current step change event, record its occurrence time and extract the voltage and current subsequences within a preset time period before and after the time. All the extracted voltage and current subsequences are stored in event order to form a perturbation response dataset.

[0009] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the steps are as follows: The dynamic response characteristics under perturbation excitation are extracted based on the collected voltage and current signals, and temperature drift correction is performed on the dynamic response characteristics in conjunction with the real-time measured battery temperature to obtain the corrected equivalent insulation resistance value. For each voltage subsequence and current subsequence in the perturbation response dataset, calculate the average voltage and average current within the steady-state window before the perturbation, and the average voltage and average current within the steady-state window after the perturbation. The difference between the average voltage before and after the disturbance is used as the voltage change, denoted as . The difference between the average current before and after the disturbance is used as the change in current, denoted as . ; If the change in current If the absolute value of the disturbance is greater than the preset disturbance validity threshold, the disturbance event is deemed valid, and the estimated instantaneous internal resistance corresponding to the disturbance event is calculated. The expression is: ; Instantaneous internal resistance estimates for all valid perturbation events A weighted average is performed, with the weights determined by the signal-to-noise ratio (SNR) of the current change for each corresponding event; the SNR is defined as the current change... The ratio to the standard deviation of the current within the disturbance window; The weighted average result is recorded as the preliminary equivalent insulation internal resistance value. ; Meanwhile, the current battery temperature is obtained by a temperature sensor placed on the surface of the backup power supply housing. ; Based on the temperature-internal resistance compensation function established in advance through experimental calibration Calculate the current temperature Corresponding compensation factor ,Right now ; Using compensation factors For the preliminary equivalent insulation internal resistance value The correction is performed to obtain the corrected equivalent insulation internal resistance value. The expression is: ; in, This is the preliminary equivalent insulation internal resistance value. This is the corrected equivalent insulation internal resistance value.

[0010] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the steps of establishing a battery connection topology model according to the battery series-parallel topology of the backup power supply in the power distribution terminal, inputting the corrected equivalent insulation resistance value and dynamic response characteristics into the sparse system identification model, and decoupling the individual insulation resistance of each single battery cell are as follows: Analyze the physical connection structure of the backup power supply of the power distribution terminal to determine the number N of individual batteries and the series and parallel relationships between the batteries; Based on this series-parallel relationship, construct the node-branch association matrix. The matrix has dimensions M×N, where M represents the number of independently measurable electrical ports; For each valid perturbation event, based on the current distribution principle, the current distribution vector of each individual cell in that event is derived. The current distribution vector satisfy ,in This represents the measured total change in current. Assume that each individual cell has an unknown internal insulation resistance, denoted as . And form a vector from the unknowns. ; According to Ohm's law, the change in port voltage caused by a perturbation event... This can be represented as a current distribution vector. Vector of internal resistance of individual insulation The inner product is expressed as: ; Collect all valid perturbation events to form a system of linear equations. ,in For all The column vector formed For all The matrix formed; Since the number of perturbation events is usually smaller than the number of individual cells, the equation set is an underdetermined system; To find the unique and reasonable solution, a sparsity prior is introduced, and the L1 norm regularized least squares method is used. The objective function is: ; in, This represents the estimated vector of individual insulation internal resistance obtained by solving the problem. This represents the system matrix consisting of the current assignment vectors of all perturbation events. This represents the observation vector composed of all measured voltage changes. This represents the regularization coefficient, used to balance the fitting error with the sparsity of the solution, and is determined by cross-validation; The obtained vector Each component represents the individual insulation resistance of the corresponding single cell.

[0011] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the following steps are taken: Continuously sampling the individual insulation resistance of each single battery cell to form an individual insulation resistance time series; calculating the information entropy and coefficient of variation of this time series within a preset sliding time window; and generating an internal resistance stability index based on the information entropy and coefficient of variation. For each individual cell, its internal insulation resistance value is recorded at fixed time intervals to form a time series; Set the sliding time window length to The sliding window moves forward with a fixed step size, and for each sliding window, the corresponding individual internal resistance sequence of the insulation is extracted. Divide the numerical range of the subsequence into Divide the sample into equal-width intervals and count the frequency of the sample occurrence in each interval. Normalize the frequency into a probability distribution ; Calculate Shannon information entropy based on probability distribution. The expression is: ; in, This represents the total number of intervals divided. Indicates the first The normalized probability of a sample appearing within each interval satisfies , The information entropy represents the internal resistance of an individual insulation within the current time window, reflecting the uncertainty or randomness of its value; At the same time, calculate the arithmetic mean of the subsequences. with standard deviation ; Using average value with standard deviation Calculate the coefficient of variation The expression is: ; in, This represents the average internal insulation resistance of individuals within the current time window. This represents the corresponding standard deviation. Indicates the coefficient of variation; Information entropy With coefficient of variation Fusion, generating internal resistance stability index The fusion method employs a weighted linear combination, with weights determined through training on historical fault data; internal resistance stability index. The expression is: ; in, This represents the preset weighting coefficient. Represents the maximum possible entropy calculated using the natural logarithm or the common logarithm, used for... Normalize, This represents the internal resistance stability index.

[0012] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the steps of obtaining communication anomaly event records of the power distribution terminal within a time period synchronized with the sliding time window, and generating multidimensional health assessment indicators by combining the internal resistance stability index, the current value of individual insulation resistance, and the communication anomaly event records, are as follows: Read the communication anomaly event records that completely overlap with the physical time period of the current sliding time window from the operation log of the power distribution terminal; Communication anomalies include interruption of the main station communication link, self-restart caused by abnormal exit of local application processes, and timeout of telemetry data reporting; The total number of communication anomalies during this time period is recorded as follows: ; Take the last sampled value of the individual insulation internal resistance sequence within the current sliding time window as the current value of the individual insulation internal resistance, denoted as . ; Current value of individual insulation internal resistance Internal resistance stability index Number of communication anomalies Combined into a three-dimensional feature vector; The three-dimensional feature vector is input into a pre-trained health status assessment model, which is a gradient boosting tree or support vector machine trained based on historical failure samples. The model outputs a value between 0 and 1 as a multidimensional health assessment indicator. ; The multidimensional health assessment indicators It comprehensively reflects the correlation between the electrical degradation of the power system and abnormal system-level behavior; the higher the value, the higher the risk of failure.

[0013] As a preferred embodiment of the power distribution terminal backup power monitoring method based on online insulation resistance detection described in this invention, the method involves generating alarm information containing the location identifier and maintenance suggestions of the corresponding individual battery when the individual insulation resistance of any single battery cell exceeds the adaptive dynamic threshold, or when the multi-dimensional health assessment indicators meet the preset early warning criteria. The alarm information is then output through a local interface or a remote communication channel. The specific steps are as follows: For each individual cell, maintain the historical time series of its individual insulation internal resistance; Based on historical time series, the adaptive early warning threshold is dynamically calculated using the moving quantile algorithm. The adaptive early warning threshold It automatically rises as the battery ages; Real-time comparison of current values ​​of individual insulation internal resistance With adaptive warning threshold ; At the same time, determine multidimensional health assessment indicators Is it greater than the preset warning threshold? ; If satisfied or If any of the following conditions are met, the individual cell is deemed to have a risk of insulation degradation. A unique location identifier is generated based on the physical installation location of each individual battery cell inside the power distribution terminal. Construct alarm information, which includes location identifier, current value of individual insulation internal resistance, internal resistance stability index, number of communication anomalies, multi-dimensional health assessment indicators and maintenance recommendation type; Maintenance recommendations include suggestions for immediate replacement, increased inspections, or immediate power outages for repairs, based on multidimensional health assessment indicators. The range of values ​​determines the value; Alarm information can be displayed locally on the LCD screen of the power distribution terminal panel, or uploaded to the power distribution automation master station operation and maintenance platform through the built-in 4G communication module and fiber optic Ethernet interface.

[0014] This invention provides a power distribution terminal backup power monitoring system based on online insulation resistance detection, comprising: The system includes a perturbation acquisition module, an internal resistance analysis module, a single-unit decoupling module, a stability analysis module, a health assessment module, and an early warning output module. The perturbation acquisition module is used to synchronously acquire the voltage and current signals of the backup power supply port when the power distribution terminal is operating normally, identify natural perturbation events and extract the corresponding sub-sequences to form a perturbation response dataset. The internal resistance analysis module is used to extract dynamic response features from the perturbation response dataset, calculate the instantaneous internal resistance estimate, and obtain the corrected equivalent insulation internal resistance value by weighted averaging and temperature compensation. The single-cell decoupling module is used to construct a system matrix based on the battery series-parallel topology and use a sparse system identification model to decouple the individual insulation resistance of each single cell from the total port response. The stability analysis module is used to calculate the information entropy and coefficient of variation of the individual insulation internal resistance time series within a sliding window, and to fuse them to generate an internal resistance stability index. The health assessment module is used to generate multidimensional health assessment indicators by combining the current value of individual insulation internal resistance, internal resistance stability index and synchronous communication abnormal event records through a pre-trained model. The warning output module is used to generate alarm information containing battery location identification and maintenance suggestions based on adaptive dynamic thresholds or multi-dimensional health assessment indicators, and output it through local or remote channels.

[0015] The beneficial effects of this invention are as follows: By utilizing the inherent natural perturbation components in the operation of the power distribution terminal, it achieves seamless and uninterrupted online identification of insulation resistance, breaking through the traditional detection paradigm that requires active injection of excitation signals. Combining the battery topology and sparse system identification model, it achieves accurate decoupling of the individual insulation resistance of a single battery cell from the total port response. By introducing information entropy and coefficient of variation to construct an internal resistance stability index, and integrating communication anomaly events to form a multi-dimensional health assessment index, it effectively improves the identification accuracy and early warning reliability of early insulation degradation and sudden faults. It solves the problems of existing technologies being unable to locate degraded cells, being susceptible to temperature interference, and having a high false alarm rate, providing highly available, intelligent, and maintainable online monitoring capabilities for the backup power supply of the power distribution terminal. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the backup power monitoring method for power distribution terminals based on online detection of insulation internal resistance in the embodiments.

[0018] Figure 2 This is a schematic diagram of a power distribution terminal backup power monitoring system based on online insulation resistance detection in the embodiment. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0021] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0022] Example 1: Reference Figure 1 and Figure 2 This embodiment provides a method for monitoring backup power supply of a power distribution terminal based on online detection of insulation internal resistance, including the following steps: S1. Collect the voltage and current signals of the backup power port of the power distribution terminal under normal operating conditions. The voltage and current signals contain natural perturbation components generated by load changes or the start and stop of the communication module.

[0023] Furthermore, the high-speed data acquisition module inside the power distribution terminal is activated to synchronously acquire the terminal voltage signal and loop current signal between the positive and negative poles of the backup power supply at a fixed sampling frequency without interrupting the terminal's business functions; the acquired voltage and current signals are timestamped to form a synchronous timing data stream.

[0024] Scan the synchronous timing data stream to identify current step change events caused by periodic wake-up of the communication module, reversal of remote signaling status, or external load access; for each identified current step change event, record its occurrence time and extract voltage and current subsequences within a preset time period before and after the time; store all extracted voltage and current subsequences in event order to form a perturbation response dataset.

[0025] It should be noted that by utilizing the inherent load changes or natural disturbances generated by the start and stop of communication modules during the normal operation of the power distribution terminal as excitation sources, no additional test signals or interruption of terminal services are required, achieving truly seamless online detection. This avoids interference with the stable operation of the power distribution terminal, reduces hardware costs and system complexity, and provides a feasible basis for large-scale deployment.

[0026] Specifically, the natural perturbation component relied upon in this step is not random noise, but rather the load steps that inevitably occur when the distribution terminal periodically performs telemetry and telesignaling tasks under the standard IEC60870-5-101 / 104 or DL / T634.5104 communication protocols. Its amplitude and frequency are repeatable and observable. The high-speed data acquisition module adopts a dual-channel synchronous ADC architecture with a sampling rate of no less than 2kHz, sufficient to capture microsecond-level current jumps and ensure that perturbation events are not aliased. Timestamp alignment is achieved through hardware triggering or the PTP precision clock protocol, ensuring that voltage and current signals are aligned within microsecond-level synchronization accuracy, laying the data foundation for subsequent dynamic response feature extraction. The design requires no additional dedicated excitation circuit, is fully compatible with existing distribution terminal hardware platforms, and has engineering implementation capabilities.

[0027] S2. Based on the collected voltage and current signals, extract the dynamic response characteristics under perturbation excitation, and combine the real-time measured battery temperature to perform temperature drift correction on the dynamic response characteristics to obtain the corrected equivalent insulation resistance value.

[0028] Furthermore, for each voltage and current subsequence in the perturbation response dataset, the average voltage and current values ​​within the steady-state window before and after the perturbation are calculated. The difference between the average voltage values ​​before and after the perturbation is used as the voltage change, denoted as... The difference between the average current before and after the disturbance is used as the change in current, denoted as . If the change in current If the absolute value of the disturbance is greater than the preset disturbance validity threshold, the disturbance event is deemed valid, and the estimated instantaneous internal resistance corresponding to the disturbance event is calculated. The expression is: ; Instantaneous internal resistance estimates for all valid perturbation events A weighted average is performed, with the weights determined by the signal-to-noise ratio (SNR) of the current change for each corresponding event; the SNR is defined as the current change... The ratio to the standard deviation of the current within the disturbance window.

[0029] The weighted average result is recorded as the preliminary equivalent insulation internal resistance value. Simultaneously, the current battery temperature is obtained through a temperature sensor located on the surface of the backup power supply casing. Based on the temperature-internal resistance compensation function established in advance through experimental calibration. Calculate the current temperature Corresponding compensation factor ,Right now ; Utilizing compensation factors For the preliminary equivalent insulation internal resistance value The correction is performed to obtain the corrected equivalent insulation internal resistance value. The expression is: ; in, This is the preliminary equivalent insulation internal resistance value. This is the corrected equivalent insulation internal resistance value.

[0030] It should be noted that the instantaneous internal resistance is calculated by using the mean difference of the steady-state window before and after the disturbance, and a signal-to-noise ratio weighted averaging mechanism is introduced to effectively suppress the influence of noise interference on the internal resistance estimation. At the same time, real-time correction is performed by combining the temperature-internal resistance compensation function based on experimental calibration, which eliminates misjudgment caused by environmental temperature drift and greatly improves the accuracy and robustness of the equivalent insulation internal resistance value.

[0031] Specifically, this step uses the difference between the steady-state window mean before and after the disturbance, rather than the instantaneous derivative, because the backup power supply has a low output impedance and the disturbance response has an extremely fast rise time (typically <10ms). Direct differentiation is easily affected by quantization noise amplification. The steady-state window averaging can effectively smooth high-frequency interference while retaining a sufficiently large mean. and Used for internal resistance calculation; the signal-to-noise ratio weighting mechanism originates from the Bayesian estimation idea—high signal-to-noise ratio events are closer to the real system response and should be assigned higher confidence levels; temperature compensation function. Instead of a simple linear fit, it is based on the Arrhenius electrochemical model and calibrates the nonlinear compensation curve through multiple sets of accelerated aging experiments in the range of 20℃ to 60℃. This ensures that the internal resistance correction error is less than ±3% under all operating conditions, and the design ensures that the internal resistance estimation maintains high consistency in complex field environments.

[0032] S3. Based on the battery series-parallel topology of the backup power supply in the power distribution terminal, establish a battery connection topology model, input the corrected equivalent insulation resistance value and dynamic response characteristics into the sparse system identification model, and decouple the individual insulation resistance of each single battery cell.

[0033] Furthermore, the physical connection structure of the backup power supply of the distribution terminal is analyzed to determine the number N of individual batteries and the series and parallel connections between them; based on these connections, a node-branch correlation matrix is ​​constructed. The matrix has dimensions M×N, where M represents the number of independently measurable electrical ports. For each valid perturbation event, based on the current distribution principle, the current distribution vector of each individual cell in that event is derived. Current distribution vector satisfy ,in This represents the measured total current change.

[0034] Assume that each individual cell has an unknown internal insulation resistance, denoted as . And form a vector from the unknowns. According to Ohm's law, the change in port voltage caused by a perturbation event... This can be represented as a current distribution vector. Vector of internal resistance of individual insulation The inner product is expressed as: ; Collect all valid perturbation events to form a system of linear equations. ,in For all The column vector formed For all The resulting matrix; since the number of perturbation events is usually less than the number of individual cells, the system of equations is an underdetermined system; to find a unique and reasonable solution, a sparsity prior is introduced, and the L1 norm regularized least squares method is used for the solution. The objective function is: ; in, This represents the estimated vector of individual insulation internal resistance obtained by solving the problem. This represents the system matrix consisting of the current assignment vectors of all perturbation events. This represents the observation vector composed of all measured voltage changes. The regularization coefficient, used to balance the fitting error and the sparsity of the solution, is determined by cross-validation; the resulting vector... Each component represents the individual insulation resistance of the corresponding single cell.

[0035] It should be noted that by modeling the battery physical topology as a node-branch correlation matrix and solving the underdetermined equations using sparse system identification theory, the precise decoupling of the individual insulation resistance of each cell in a series-parallel battery pack was achieved for the first time under the condition of measuring only the total port response. This breakthrough overcomes the technical bottleneck of traditional methods being unable to locate degraded cells and provides key data support for refined operation and maintenance.

[0036] In particular, the effectiveness of the sparse system identification model is based on the prior assumption that only a few cells in the battery pack deteriorate, which is consistent with the statistical law of lithium battery aging—insulation degradation is usually caused by local defects and has sparsity. The construction of the node-branch correlation matrix A strictly follows Kirchhoff's law, and its rank is equal to the number of independent loops, ensuring the physical rationality of the current allocation vector i. The sparsity of the solution induced by the L1 regularization term ‖x‖1 enables the optimization process to automatically suppress the internal resistance estimation bias of non-deteriorated batteries. The regularization coefficient λ is automatically selected on the historical dataset through K-fold cross-validation to avoid overfitting. The method achieves decoupling of the internal resistance of cells for the first time under the condition of no cell voltage monitoring, breaking through the hardware limitation of traditional BMS relying on cell sampling lines, and is suitable for low-cost power distribution terminal scenarios.

[0037] S4. Continuously sample the individual insulation resistance of each cell to form an individual insulation resistance time series. Calculate the information entropy and coefficient of variation of the time series within a preset sliding time window, and generate an internal resistance stability index based on the information entropy and coefficient of variation.

[0038] Furthermore, for each individual cell, its internal insulation resistance value is recorded at fixed time intervals to form a time series; the sliding time window length is set to... Slide forward with a fixed step size, and for each sliding window, extract the corresponding individual insulation internal resistance sub-sequence; divide the numerical range of the sub-sequence into... Divide the data into equal-width intervals and count the frequency of samples within each interval; then normalize the frequencies to a probability distribution. Based on probability distribution, calculate Shannon information entropy. The expression is: ; in, This represents the total number of intervals divided. Indicates the first The normalized probability of a sample appearing within each interval satisfies , It represents the information entropy of an individual's insulation internal resistance within the current time window, reflecting the uncertainty or randomness of its value.

[0039] At the same time, calculate the arithmetic mean of the subsequences. with standard deviation Using average value with standard deviation Calculate the coefficient of variation The expression is: ; in, This represents the average internal insulation resistance of individuals within the current time window. This represents the corresponding standard deviation. Represents the coefficient of variation; information entropy With coefficient of variation Fusion, generating internal resistance stability index The fusion method employs a weighted linear combination, with weights determined through training on historical fault data; internal resistance stability index. The expression is: ; in, This represents the preset weighting coefficient. Represents the maximum possible entropy calculated using the natural logarithm or the common logarithm, used for... Normalize, This represents the internal resistance stability index.

[0040] It should be noted that the introduction of information entropy and coefficient of variation to construct the internal resistance stability index not only focuses on the change of the absolute value of internal resistance, but also delves deeper into the fluctuation regularity and randomness of its time series. This can effectively distinguish between the slow drift caused by normal aging and the sudden change or oscillation caused by early faults, effectively improving the sensitivity and discrimination ability to the early signs of insulation degradation.

[0041] In particular, information entropy The distribution dispersion of the internal resistance sequence reflects the slow, monotonically increasing internal resistance of a normal battery, with the probability concentrated in the high-value range. The probability distribution is relatively small; while early faults often manifest as internal resistance oscillations or sudden jumps, resulting in a flat probability distribution. Significantly increased; coefficient of variation This eliminates the influence of dimensions and highlights the relative intensity of fluctuations; the fusion of these two factors can distinguish between two risk modes: high internal resistance but stable (late aging stage) and medium internal resistance but volatile (early deterioration); weighting Fault samples are used to determine the cause, such as in lead-acid batteries. A value of 0.6 is chosen to emphasize information entropy, while 0.4 is chosen in lithium batteries to emphasize the coefficient of variation; sliding window length The timeframe is set to 24 hours, balancing short-term anomaly detection with long-term trend smoothing. The index reflects the dynamic nature of degradation more effectively than a single threshold criterion.

[0042] S5. Obtain communication anomaly event records of the power distribution terminal within the time period synchronized with the sliding time window, and generate multidimensional health assessment indicators by combining the internal resistance stability index, the current value of individual insulation internal resistance, and the communication anomaly event records.

[0043] Furthermore, in the operation log of the power distribution terminal, records of communication anomaly events that completely overlap with the current sliding time window's physical time period are read; these communication anomaly events include main station communication link interruption, self-restart caused by abnormal exit of local application processes, and telemetry data reporting timeout; the total number of communication anomaly events occurring within this time period is counted and recorded as follows. Take the last sampled value of the individual insulation internal resistance sequence within the current sliding time window, and use it as the current value of the individual insulation internal resistance, denoted as . ;The current value of the individual insulation internal resistance Internal resistance stability index Number of communication anomalies The features are combined into a three-dimensional feature vector; this three-dimensional feature vector is then input into a pre-trained health status assessment model, which is either a gradient boosting tree or a support vector machine trained based on historical fault samples; the model outputs a value between 0 and 1, serving as a multidimensional health assessment indicator. Multidimensional health assessment indicators It comprehensively reflects the correlation between the electrical degradation of the power system and abnormal system-level behavior; the higher the value, the higher the risk of failure.

[0044] It should be noted that by spatiotemporally integrating power supply electrical parameters (internal resistance, stability index) with system-level behavioral anomalies (communication anomalies) to construct multidimensional health assessment indicators, a leap from judging a single electrical quantity to electrical-communication collaborative diagnosis has been achieved. This effectively overcomes false alarms or missed alarms caused by fluctuations in a single indicator, making the health status assessment closer to the actual fault evolution mechanism.

[0045] In particular, there is a strong causal relationship between communication anomalies and power supply degradation: when the internal resistance of the backup power supply increases, causing a drop in the supply voltage, the terminal main control chip may reset, triggering a self-restart or communication interruption; conversely, frequent communication anomalies may also exacerbate power supply load shocks and accelerate degradation. This step achieves spatiotemporal coupling of telecommunications data through strict alignment of time windows, avoiding the introduction of spurious correlations by asynchronous associations; the three-dimensional feature vectors are input into a gradient boosting tree (such as XGBoost) after Z-score standardization, which can automatically learn the nonlinear interactions between features (such as "high"). +High "The risk of a portfolio is far greater than that of a single high-risk portfolio." The training samples cover operational data from hundreds of terminals under different climates and loads, ensuring the model's generalization ability. This multi-source fusion mechanism can effectively improve the specificity of early warnings.

[0046] S6. When the internal insulation resistance of any single cell exceeds the adaptive dynamic threshold, or when the multi-dimensional health assessment indicators meet the preset warning criteria, an alarm message containing the location identifier of the corresponding single cell and maintenance suggestions is generated and output through the local interface or remote communication channel.

[0047] Furthermore, for each individual cell, maintain the historical time series of its individual insulation resistance; Based on historical time series, the adaptive early warning threshold is dynamically calculated using the moving quantile algorithm. Adaptive early warning threshold Automatically rises as the battery ages; compares the current value of individual insulation internal resistance in real time. With adaptive warning threshold Simultaneously, assess multidimensional health indicators. Is it greater than the preset warning threshold? If satisfied or If any of the following conditions are met, the individual battery cell is determined to have a risk of insulation degradation. A unique location identifier is generated based on the physical installation location of the individual battery cell within the power distribution terminal. An alarm message is constructed, including the location identifier, the current value of the individual battery cell's internal insulation resistance, the internal resistance stability index, the number of communication anomalies, multi-dimensional health assessment indicators, and the type of maintenance recommendation. The maintenance recommendation type includes recommendations for immediate replacement, increased inspection, or immediate power-off maintenance, based on the multi-dimensional health assessment indicators. The numerical range is determined; alarm information is displayed locally on the LCD screen of the power distribution terminal panel, or uploaded to the power distribution automation master station operation and maintenance platform through the built-in 4G communication module and fiber optic Ethernet interface.

[0048] It should be noted that the sliding quantile algorithm is used to dynamically generate adaptive warning thresholds, which automatically adjust the thresholds according to the battery aging trend, avoiding the problem of fixed thresholds being too sensitive in the early stage of life or failing in the late stage. At the same time, combined with multi-dimensional health assessment indicators, graded warnings and differentiated maintenance suggestions are implemented, realizing the upgrade from no alarms to precise decision-making, which greatly improves operation and maintenance efficiency and power supply reliability.

[0049] Specifically, the adaptive dynamic threshold Rth uses a sliding 90th percentile instead of a fixed multiple because the battery's internal resistance increases non-linearly with the number of cycles, increasing slowly in the early stages and then sharply in the later stages. The sliding quantile can automatically track this trend, avoiding frequent alarms due to excessively low thresholds in the early stages, or missed alarms due to threshold lag in the later stages; multi-dimensional health assessment indicators The tiered strategy is based on the principle of balancing operation and maintenance costs and risks: For values ​​∈(0.3,0.6], it is recommended to strengthen inspections. For values ​​∈(0.6,0.85], immediate replacement is recommended; for D>0.85, immediate power cut-off is required. This threshold is jointly calibrated through Monte Carlo simulation and on-site fault backtracking. Alarm information contains structured fields, supporting automatic parsing and work order generation by the master station, realizing a closed loop of detection-diagnosis-decision-execution. This design transforms passive response into proactive predictive maintenance, effectively improving the availability of the power distribution automation system.

[0050] Example 2: This embodiment provides a power distribution terminal backup power monitoring system based on online insulation resistance detection, including: The system includes a perturbation acquisition module, an internal resistance analysis module, a single-unit decoupling module, a stability analysis module, a health assessment module, and an early warning output module. The perturbation acquisition module is used to synchronously acquire the voltage and current signals of the backup power supply port when the power distribution terminal is operating normally, identify natural perturbation events and extract the corresponding sub-sequences to form a perturbation response dataset. The internal resistance analysis module is used to extract dynamic response features from the perturbation response dataset, calculate the instantaneous internal resistance estimate, and obtain the corrected equivalent insulation internal resistance value by weighted averaging and temperature compensation. The single-cell decoupling module is used to construct a system matrix based on the battery series-parallel topology and use a sparse system identification model to decouple the individual insulation resistance of each single cell from the total port response. The stability analysis module is used to calculate the information entropy and coefficient of variation of individual insulation internal resistance time series within a sliding window, and then fuse them to generate an internal resistance stability index. The health assessment module is used to generate multidimensional health assessment indicators by combining the current value of individual insulation internal resistance, internal resistance stability index and synchronous communication abnormal event records through a pre-trained model. The early warning output module is used to generate alarm information containing battery location identification and maintenance suggestions based on adaptive dynamic thresholds or multi-dimensional health assessment indicators, and output it through local or remote channels.

[0051] In summary, this invention achieves seamless and uninterrupted online identification of insulation resistance by utilizing the inherent natural perturbation components in the operation of the power distribution terminal. This breaks through the traditional detection paradigm that requires active injection of excitation signals. By combining battery topology and sparse system identification models, it accurately decouples the individual insulation resistance of individual cells from the total port response. By introducing information entropy and coefficient of variation to construct an internal resistance stability index, and by integrating communication anomaly events to form a multi-dimensional health assessment index, it effectively improves the identification accuracy and early warning reliability of early insulation degradation and sudden faults. It solves the problems of existing technologies being unable to locate degraded cells, being susceptible to temperature interference, and having a high false alarm rate, providing highly available, intelligent, and maintainable online monitoring capabilities for power distribution terminal backup power supplies.

[0052] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring backup power supply of a distribution terminal based on online insulation resistance detection, characterized in that: include: The voltage and current signals of the backup power port of the power distribution terminal are collected under normal operating conditions. The voltage and current signals include natural perturbation components generated by load changes or the start and stop of the communication module. Based on the collected voltage and current signals, the dynamic response characteristics under perturbation excitation are extracted, and the dynamic response characteristics are corrected for temperature drift by combining the real-time measured battery temperature, so as to obtain the corrected equivalent insulation internal resistance value. A battery connection topology model is established based on the battery series-parallel topology of the backup power supply in the power distribution terminal. The corrected equivalent insulation resistance and dynamic response characteristics are input into the sparse system identification model to decouple the individual insulation resistance of each cell. The individual insulation resistance of each cell is continuously sampled to form an individual insulation resistance time series. The information entropy and coefficient of variation of the time series are calculated within a sliding time window of a preset length. The internal resistance stability index is generated based on the information entropy and coefficient of variation. Obtain communication anomaly event records of the power distribution terminal within the time period synchronized with the sliding time window, and generate multidimensional health assessment indicators by combining the internal resistance stability index, the current value of individual insulation internal resistance, and the communication anomaly event records. When the internal insulation resistance of any single cell exceeds the adaptive dynamic threshold, or when the multidimensional health assessment indicators meet the preset warning criteria, an alarm message containing the location identifier of the corresponding single cell and maintenance suggestions is generated and output through the local interface or remote communication channel.

2. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 1, characterized in that: The specific steps for collecting the voltage and current signals of the backup power port of the power distribution terminal under normal operating conditions are as follows: The high-speed data acquisition module inside the power distribution terminal is activated to synchronously acquire the terminal voltage signal and loop current signal between the positive and negative poles of the backup power supply at a fixed sampling frequency without interrupting the terminal's business functions. The acquired voltage and current signals are timestamped to form a synchronous timing data stream; Scan the synchronous timing data stream to identify current step change events caused by periodic wake-up of the communication module, telemetry status flipping, or external load access; For each identified current step change event, record its occurrence time and extract the voltage and current subsequences within a preset time period before and after the time. All the extracted voltage and current subsequences are stored in event order to form a perturbation response dataset.

3. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 2, characterized in that: The dynamic response characteristics under perturbation excitation are extracted based on the collected voltage and current signals, and temperature drift correction is performed on the dynamic response characteristics in combination with the real-time measured battery temperature to obtain the corrected equivalent insulation resistance value. The specific steps are as follows: For each voltage subsequence and current subsequence in the perturbation response dataset, calculate the average voltage and average current within the steady-state window before the perturbation, and the average voltage and average current within the steady-state window after the perturbation. The difference between the average voltage before and after the disturbance is used as the voltage change, denoted as . The difference between the average current before and after the disturbance is used as the change in current, denoted as . ; If the change in current If the absolute value of the disturbance is greater than the preset disturbance validity threshold, the disturbance event is deemed valid, and the estimated instantaneous internal resistance corresponding to the disturbance event is calculated. The expression is: ； Instantaneous internal resistance estimates for all valid perturbation events A weighted average is performed, with the weights determined by the signal-to-noise ratio (SNR) of the current change for each corresponding event; the SNR is defined as the current change... The ratio to the standard deviation of the current within the disturbance window; The weighted average result is recorded as the preliminary equivalent insulation internal resistance value. ; Meanwhile, the current battery temperature is obtained by a temperature sensor placed on the surface of the backup power supply housing. ; Based on the temperature-internal resistance compensation function established in advance through experimental calibration Calculate the current temperature Corresponding compensation factor ,Right now ; Using compensation factors Preliminary equivalent insulation internal resistance value The correction is performed to obtain the corrected equivalent insulation internal resistance value. The expression is: ； in, This is the preliminary equivalent insulation internal resistance value. This is the corrected equivalent insulation internal resistance value.

4. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 3, characterized in that: The steps are as follows: First, a battery connection topology model is established based on the battery series-parallel topology of the backup power supply in the power distribution terminal. Then, the corrected equivalent insulation resistance and dynamic response characteristics are input into the sparse system identification model to decouple the individual insulation resistance of each battery cell. Analyze the physical connection structure of the backup power supply of the power distribution terminal to determine the number N of individual batteries and the series and parallel relationships between the batteries; Based on this series-parallel relationship, construct the node-branch association matrix. The matrix has dimensions M×N, where M represents the number of independently measurable electrical ports; For each valid perturbation event, based on the current distribution principle, the current distribution vector of each individual cell in that event is derived. The current distribution vector satisfy ,in This represents the measured total change in current. Assume that each individual cell has an unknown internal insulation resistance, denoted as . And form a vector from the unknowns. ; According to Ohm's law, the change in port voltage caused by a perturbation event... This can be represented as a current distribution vector. Vector of internal resistance of individual insulation The inner product is expressed as: ； Collect all valid perturbation events to form a system of linear equations. ,in For all The column vector formed For all The matrix formed; Since the number of perturbation events is usually smaller than the number of individual cells, the equation set is an underdetermined system; To find the unique and reasonable solution, a sparsity prior is introduced, and the L1 norm regularized least squares method is used. The objective function is: ； in, This represents the estimated vector of individual insulation internal resistance obtained by solving the problem. This represents the system matrix consisting of the current assignment vectors of all perturbation events. This represents the observation vector composed of all measured voltage changes. This represents the regularization coefficient, used to balance the fitting error with the sparsity of the solution, and is determined by cross-validation; The obtained vector Each component represents the individual insulation resistance of the corresponding single cell.

5. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 4, characterized in that: The process involves continuously sampling the individual insulation resistance of each battery cell to form an individual insulation resistance time series. Within a preset sliding time window, the information entropy and coefficient of variation of this time series are calculated. Based on the information entropy and coefficient of variation, an internal resistance stability index is generated. The specific steps are as follows: For each individual cell, its internal insulation resistance value is recorded at fixed time intervals to form a time series; Set the sliding time window length to The sliding window moves forward with a fixed step size, and for each sliding window, the corresponding individual internal resistance sequence of the insulation is extracted. Divide the numerical range of the subsequence into Divide the sample into equal-width intervals and count the frequency of the sample occurrence in each interval. Normalize the frequency into a probability distribution ; Calculate Shannon information entropy based on probability distribution. The expression is: ； in, This represents the total number of intervals divided. Indicates the first The normalized probability of a sample appearing within each interval satisfies , The information entropy represents the internal resistance of an individual insulation within the current time window, reflecting the uncertainty or randomness of its value; At the same time, calculate the arithmetic mean of the subsequences. with standard deviation ; Using average value with standard deviation Calculate the coefficient of variation The expression is: ； in, This represents the average internal insulation resistance of individuals within the current time window. This represents the corresponding standard deviation. Indicates the coefficient of variation; Information entropy With coefficient of variation Fusion, generating internal resistance stability index The fusion method employs a weighted linear combination, with weights determined through training on historical fault data; internal resistance stability index. The expression is: ； in, This represents the preset weighting coefficient. Represents the maximum possible entropy calculated using the natural logarithm or the common logarithm, used for... Normalize, This represents the internal resistance stability index.

6. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 5, characterized in that: The steps for obtaining communication anomaly event records of the power distribution terminal within a time period synchronized with the sliding time window, and combining these records with the internal resistance stability index, the current value of the individual insulation internal resistance, and the communication anomaly event records to generate a multidimensional health assessment index are as follows: Read the communication anomaly event records that completely overlap with the physical time period of the current sliding time window from the operation log of the power distribution terminal; Communication anomalies include interruption of the main station communication link, self-restart caused by abnormal exit of local application processes, and timeout of telemetry data reporting; The total number of communication anomalies during this time period is recorded as follows: ; Take the last sampled value of the individual insulation internal resistance sequence within the current sliding time window as the current value of the individual insulation internal resistance, denoted as . ; Current value of individual insulation internal resistance Internal resistance stability index Number of communication anomalies Combined into a three-dimensional feature vector; The three-dimensional feature vector is input into a pre-trained health status assessment model, which is a gradient boosting tree or support vector machine trained based on historical failure samples. The model outputs a value between 0 and 1 as a multidimensional health assessment indicator. ; The multidimensional health assessment indicators It comprehensively reflects the correlation between the electrical degradation of the power system and abnormal system-level behavior; the higher the value, the higher the risk of failure.

7. The method for monitoring backup power supply of distribution terminal based on online insulation resistance detection as described in claim 6, characterized in that: When the internal insulation resistance of any single cell exceeds the adaptive dynamic threshold, or when the multi-dimensional health assessment indicators meet the preset warning criteria, an alarm message containing the location identifier of the corresponding single cell and maintenance suggestions is generated and output through a local interface or remote communication channel. The specific steps are as follows: For each individual cell, maintain the historical time series of its individual insulation internal resistance; Based on historical time series, the adaptive early warning threshold is dynamically calculated using the moving quantile algorithm. The adaptive early warning threshold It automatically rises as the battery ages; Real-time comparison of current values ​​of individual insulation internal resistance With adaptive warning threshold ; At the same time, determine multidimensional health assessment indicators Is it greater than the preset warning threshold? ; If satisfied or If any of the following conditions are met, the individual cell is deemed to have a risk of insulation degradation. A unique location identifier is generated based on the physical installation location of each individual battery cell inside the power distribution terminal. Construct alarm information, which includes location identifier, current value of individual insulation internal resistance, internal resistance stability index, number of communication anomalies, multi-dimensional health assessment indicators and maintenance recommendation type; Maintenance recommendations include suggestions for immediate replacement, increased inspections, or immediate power outages for repairs, based on multidimensional health assessment indicators. The range of values ​​determines the value; Alarm information can be displayed locally on the LCD screen of the power distribution terminal panel, or uploaded to the power distribution automation master station operation and maintenance platform through the built-in 4G communication module and fiber optic Ethernet interface.

8. A power distribution terminal backup power supply monitoring system based on online insulation resistance detection, based on the power distribution terminal backup power supply monitoring method based on online insulation resistance detection according to any one of claims 1 to 7, characterized in that: include: The system includes a perturbation acquisition module, an internal resistance analysis module, a single-unit decoupling module, a stability analysis module, a health assessment module, and an early warning output module. The perturbation acquisition module is used to synchronously acquire the voltage and current signals of the backup power supply port when the power distribution terminal is operating normally, identify natural perturbation events and extract the corresponding sub-sequences to form a perturbation response dataset. The internal resistance analysis module is used to extract dynamic response features from the perturbation response dataset, calculate the instantaneous internal resistance estimate, and obtain the corrected equivalent insulation internal resistance value by weighted averaging and temperature compensation. The single-cell decoupling module is used to construct a system matrix based on the battery series-parallel topology and use a sparse system identification model to decouple the individual insulation resistance of each single cell from the total port response. The stability analysis module is used to calculate the information entropy and coefficient of variation of the individual insulation internal resistance time series within a sliding window, and to fuse them to generate an internal resistance stability index. The health assessment module is used to generate multidimensional health assessment indicators by combining the current value of individual insulation internal resistance, internal resistance stability index and synchronous communication abnormal event records through a pre-trained model. The warning output module is used to generate alarm information containing battery location identification and maintenance suggestions based on adaptive dynamic thresholds or multi-dimensional health assessment indicators, and output it through local or remote channels.

Images