A battery insulation abnormal discharge monitoring method based on spatial electric field

Abstract

The application belongs to the technical field of electrical equipment monitoring, and discloses a battery insulation abnormal discharge monitoring method based on a space electric field. The method comprises the following steps: a joint baseline model is established by acquiring space electric field and temperature configuration information around a battery module; then, running signals are extracted for correction and comparison to generate a disturbance event list; subsequently, the list is subjected to time sequence and space feature analysis and stage determination to generate a risk level record; finally, the record is matched with a graded disposal instruction and the baseline model is updated to form a closed-loop control link. Through multi-dimensional fusion of space electric field and temperature signals, the application realizes early and accurate identification and graded early warning of battery insulation degradation and local discharge risk, and significantly improves the operation safety and state diagnosis capability of the energy storage system.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment monitoring technology, and particularly relates to a method for monitoring abnormal discharge of battery insulation based on a spatial electric field. Background Technology

[0002] In the field of electrical equipment monitoring technology, existing solutions mainly rely on methods such as high-voltage circuit insulation resistance detection at the whole pack or vehicle level, battery pack housing temperature rise threshold alarms, safety valve opening status reporting, and charging circuit ground leakage current monitoring to construct a safety alarm chain. These methods have significant limitations:

[0003] Delayed response and lack of localization: Because existing solutions rely on system-level macroscopic electrical parameters or overall temperature changes, they are not sensitive to early, localized, and weak insulation abnormalities within battery modules. Alarms are often only triggered when insulation has severely deteriorated, even leading to localized overheating or visible damage, thus missing the optimal window for warning and intervention. Furthermore, these methods completely lack spatial fault location capabilities, failing to pinpoint which battery module or part of the module is malfunctioning, significantly increasing the difficulty and cost of maintenance and troubleshooting.

[0004] Coarse-grained status assessment: Existing methods mostly rely on fixed thresholds for a single sensing quantity to make a binary judgment of "normal / abnormal," lacking the ability to continuously and finely classify the insulation status. This means that the system cannot provide any early warning information when the battery operating status shows slight abnormalities but has not reached the alarm threshold, and cannot distinguish different stages of insulation degradation (such as early weakening, continuous degradation, and precursors to breakdown), thus failing to support preventive maintenance and accurate power scheduling decisions. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for monitoring abnormal discharge of battery insulation based on a spatial electric field, comprising:

[0006] S100: Acquire configuration data of the space electric field sensing array, configuration data of the temperature measurement node, and location information of the target battery module, and generate a joint baseline model of space electric field and temperature after standardization processing;

[0007] S200: Extract the operating signals from the working probe, reference probe, and temperature measurement node; after signal correction and deviation comparison processing, generate a list of candidate disturbance events.

[0008] S300: Perform temporal and spatial feature analysis on the list of disturbance candidate events, determine the execution phase, and generate a risk level record;

[0009] S400: Based on the risk level record, execute graded disposal instructions matching and format encapsulation, and synchronously update the joint baseline model of spatial electric field and temperature to form a closed-loop monitoring and control link.

[0010] Furthermore, the configuration data of the space electric field sensing array, the configuration data of the temperature measurement nodes, and the location information of the target battery module include:

[0011] The configuration data of the space electric field sensing array includes the arrangement information and channel identification information of the electric field sensing units. The electric field sensing units are divided into working probes and reference probes. The working probes are non-contact electric field acquisition units that are close to the high potential difference area, the pole area and the seam area of ​​the battery casing. The reference probes are non-contact electric field acquisition units that are installed far away from the high potential difference area and the area with low thermal gradient.

[0012] The arrangement and channel identification information of the electric field sensing unit, wherein the electric field sensing unit is divided into a working probe close to the high potential difference area of ​​the battery and a reference probe far away from the area;

[0013] The temperature measurement node configuration data includes the installation location, numbering scheme, and reading range of the temperature acquisition units arranged close to the outer surface of the target battery module housing or the inner wall of the cavity near the battery module.

[0014] The temperature measurement node configuration data includes the installation location, number, and reading range of the temperature acquisition unit;

[0015] The target battery module location information includes the three-dimensional installation reference position, installation orientation, fastening boundary, and accessible maintenance surface information of the target battery module in the cabinet, energy storage compartment, or vehicle battery compartment.

[0016] The target battery module location information includes its three-dimensional installation reference, orientation, fastening boundary, and maintenance surface information.

[0017] Furthermore, the temporal and spatial characteristics of the list of disturbance candidate events are analyzed, and the process of determination during the execution phase includes:

[0018] The coordinates of each event entry in the list of disturbance candidate events are bound to the region as the region primary key. The time tags are arranged in chronological order. Under the same region, it is determined whether there is a repeated triggering trend and a persistence marker is generated.

[0019] Using the coordinate-bound region as an index, the event labels are arranged in chronological order to determine the recurring triggering trend and generate persistent markers;

[0020] The persistence flag writes the number of repetitions, the interval between adjacent triggers, and the span of the trigger distribution when the identified region repeats the triggering behavior, in order to indicate the temporal persistence characteristics of the local differential electric field disturbance;

[0021] The persistent marker includes the number of repetitions, the trigger interval, and the distribution span.

[0022] Furthermore, the process of determining and generating a risk level record during the execution phase includes:

[0023] Locate multiple differential component fragment indices with the same probe number from the list of candidate disturbance events, combine them with the original time-series sampling content of the synchronously acquired data packets during operation, reconstruct the trajectory of the local differential electric field disturbance amplitude change, and generate time feature annotation results.

[0024] Furthermore, the process of generating time feature annotation results includes:

[0025] The time feature annotation results record the number of recurrences of each coordinate-bound region, the peak level of the disturbance amplitude, the duration of the disturbance, the start and end time labels of the disturbance, and the persistence marker information.

[0026] Furthermore, the process of determining and generating risk level records during the execution phase also includes:

[0027] Based on the coordinate binding relationship in the array deployment record, the set of working probes located on the same shell surface, shell corner, or pole neighborhood is determined and spatial feature annotation results are formed.

[0028] Furthermore, the process of determining and generating risk level records during the execution phase also includes:

[0029] The spatial feature annotation results are a summary description of the physically adjacent regions, recording the set of working probe numbers, persistent markers, amplitude evolution trajectory summaries, and spatial adjacency registration information for the region.

[0030] Furthermore, it also includes:

[0031] The spatial feature annotation results are compared in parallel with the temperature measurement node signals. Based on the continuous marking and temperature evolution trend, early weakening tendency, insulation degradation or suspected local breakdown precursor areas are determined.

[0032] Furthermore, the temperature measurement node signal is a time series of temperature readings collected by a temperature sensing unit attached to the outer wall of the target battery module housing or the inner wall of an adjacent cavity, and it maintains a coordinate binding relationship with the spatial feature annotation results.

[0033] Furthermore, the risk level record includes:

[0034] Each risk level record includes regional coordinates, cluster grouping identifiers, persistence markers, a summary of amplitude evolution trajectory, correlation results of temperature measurement node signals, and stage judgment conclusions.

[0035] The key innovations of this invention include:

[0036] (1) Obtain the configuration data of the space electric field sensing array, the configuration data of the temperature measurement node and the location information of the target battery module, and perform probe fixed position registration, number registration, coordinate binding, synchronous sampling test, background noise extraction, initial temperature sampling, multi-channel time alignment, common mode noise subtraction and temperature value binding processing to generate a space electric field-temperature joint baseline model, forming a joint reference structure of the space electric field sensing array and temperature measurement node around the target battery module, instead of relying on a single channel signal or a single threshold.

[0037] (2) Based on the space electric field-temperature joint baseline model, the synchronous acquisition data packets during operation are obtained, and the working probe signal, reference probe signal and temperature measurement node signal are extracted from the synchronous acquisition data packets during operation. Filtering and denoising, time synchronization and differential calculation are performed, and channel-by-channel deviation comparison processing is performed to obtain a list of disturbance candidate events. Subsequently, the list of disturbance candidate events and the synchronous acquisition data packets during operation are obtained. The list of disturbance candidate events is screened for repeated occurrences at the same location, amplitude evolution statistics and persistence marking are performed, and adjacent working probe records are clustered and spatial adjacency registration is performed. The spatial feature annotation results and temperature measurement node signals are processed for stage judgment, and risk level records are generated. The time-space-temperature joint interpretation link from channel-by-channel deviation comparison processing to risk level records is constructed.

[0038] (3) Obtain risk level records, perform maintenance and inspection marking by regional coordinates, load reduction instruction template matching and isolation instruction template matching from the risk level records, and encapsulate the output format of the superior management unit and register the module-level execution status, return status registration and update the space electric field-temperature joint baseline model for the graded disposal draft, generate the graded disposal instruction set and the updated space electric field-temperature joint baseline model, so that the risk level records directly drive the graded disposal instruction set and write back the updated space electric field-temperature joint baseline model, forming a closed loop of baseline generation, disturbance identification, risk level recording, graded disposal instruction set and updated space electric field-temperature joint baseline model.

[0039] The following are its main beneficial effects:

[0040] (1) A high-precision, traceable joint reference benchmark was established: By constructing a "space electric field-temperature joint baseline model", this invention accurately binds and synchronously calibrates the space electric field, temperature, and physical location of the battery module under a unified coordinate system. This provides a dynamic and high-precision three-dimensional reference benchmark for battery status monitoring during operation, effectively overcoming the problem that traditional methods cannot identify early and weak insulation discharge signals due to the lack of effective reference when the battery has no obvious thermal failure or abnormal electrical parameters.

[0041] (2) The present invention achieves interpretable and localizable transformation from signal to risk: Through the processing link from "disturbance candidate event list" to "risk level record", the original electric field disturbance signal is transformed into a risk level conclusion for a specific battery module and a specific coordinate area by combining its temporal repeatability, spatial correlation and temperature evolution trend. This process enables early capture, precise location and risk level classification of abnormal insulation discharge events, so that the monitoring results can not only provide alarms, but also guide subsequent precise operation and maintenance.

[0042] (3) A complete "perception-analysis-decision-optimization" monitoring and control closed loop has been formed: The risk level record generated by this invention can directly drive the generation of graded disposal instructions (such as marking inspection, load reduction, and isolation), and update the baseline model simultaneously. This forms an automated closed-loop control link, enabling the monitoring system not only to detect problems, but also to trigger corresponding control actions, and to continuously optimize its own monitoring benchmark based on the new state after disposal, thereby continuously improving the operational safety and adaptive capability of the energy storage system in terms of state diagnosis. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating a battery insulation abnormal discharge monitoring method based on a spatial electric field, provided as an embodiment of this application. Detailed Implementation

[0044] Example 1: Refer to Figure 1 This is a flowchart illustrating a battery insulation abnormal discharge monitoring method based on a spatial electric field, provided by an embodiment of the present invention. The flowchart may include at least steps S100-S400:

[0045] S100: Acquire configuration data of the space electric field sensing array, configuration data of the temperature measurement node, and location information of the target battery module, and generate a joint baseline model of space electric field and temperature after standardization processing;

[0046] S200: Extract the operating signals from the working probe, reference probe, and temperature measurement node; after signal correction and deviation comparison processing, generate a list of candidate disturbance events.

[0047] S300: Perform temporal and spatial feature analysis on the list of disturbance candidate events, determine the execution phase, and generate a risk level record;

[0048] S400: Based on the risk level record, execute graded disposal instructions matching and format encapsulation, and synchronously update the joint baseline model of spatial electric field and temperature to form a closed-loop monitoring and control link.

[0049] S100: Acquire configuration data of the space electric field sensing array, configuration data of the temperature measurement node, and location information of the target battery module, and generate a joint baseline model of space electric field and temperature after standardization processing;

[0050] Specifically, the configuration data of the space electric field sensing array is first obtained. This configuration data includes the arrangement information and channel identification information of the electric field sensing units surrounding the target battery module.

[0051] The target battery module is a single battery module unit within an electrochemical energy storage system; the electric field sensing unit is divided into a working probe and a reference probe.

[0052] The working probe is a non-contact electric field acquisition unit located near the high potential difference region, electrode region, and shell seam region of the battery casing, used to record the instantaneous spatial electric field intensity fluctuations near these locations;

[0053] The reference probe is a non-contact electric field acquisition unit installed far away from areas with high potential difference and low thermal gradient, used to record the overall background field components and non-local disturbance quantities.

[0054] The temperature measurement node configuration data is acquired synchronously. The temperature measurement node configuration data includes the installation location, numbering scheme and reading range of the temperature acquisition unit arranged close to the outer surface of the target battery module housing or the inner wall of the cavity near the battery module. The temperature measurement node is used to record the ambient temperature around the battery module and the local temperature rise distribution of the battery housing.

[0055] Further, the target battery module location information is obtained. This information includes the target battery module's three-dimensional installation reference position, installation orientation, fastening boundaries, and accessible maintenance surface information within the cabinet, energy storage compartment, or vehicle battery compartment. This information is used to establish probe coordinate binding relationships later. Using the spatial electric field sensing array configuration data, temperature measurement node configuration data, and target battery module location information as input, probe fixed position registration, numbering registration, and coordinate binding are performed on each electric field sensing unit and temperature measurement node. Probe fixed position registration involves confirming the correspondence between the actual installation position and the preset installation point and recording the installation completion status. Numbering registration assigns a unique number to each working probe, reference probe, and temperature measurement node and records the mapping relationship between this number and the channel acquisition address. Coordinate binding involves mapping each numbered probe or temperature measurement node to the target battery module's shell coordinate system and writing it into the array deployment record. The array deployment record, once formed, includes the number, coordinates, pointing area attributes, and binding relationship with the target battery module for each working probe, reference probe, and temperature measurement node. This array deployment record serves as the basic input in subsequent steps, eliminating the need for repeated manual positioning.

[0056] Furthermore, based on the aforementioned array deployment records, synchronous sampling tests are performed on all registered working probes, reference probes, and temperature measurement nodes. The synchronous sampling test refers to simultaneously triggering data readings from all channels while the target battery module is in its nominal static or low-power stable operating condition, recording the spatial electric field and temperature readings at the same time. Specifically, during the synchronous sampling test, for the working probe, the instantaneous electric field strength value near the high potential difference region is extracted as the local electric field response signal; for the reference probe, the electric field strength value in a relatively distant region is extracted as the background electric field response signal; and for the temperature measurement nodes, the real-time temperature readings around the target battery module are read and the spatial location of these readings is marked. Furthermore, during the synchronous sampling test, background noise extraction is performed. This involves recording the low-amplitude, slowly varying components of all probe outputs individually, under conditions of no external disturbance injection, no high-frequency switching, and no manual touching of the wiring. This is used to identify the ambient electromagnetic field and structurally coupled electric field. Simultaneously, initial temperature sampling is performed. This involves continuously extracting the readings of the temperature measurement nodes within a preset short time window to obtain a snapshot of the temperature distribution of the target battery module under static or stable operating conditions. Through the above synchronous sampling test, background noise extraction, and initial temperature sampling, the electric field readings, background noise readings, and temperature readings at the same moment are encapsulated into an initial synchronous sampling dataset. This initial synchronous sampling dataset records the number, corresponding coordinates, instantaneous reading, and time stamp of each working probe, reference probe, and temperature measurement node. This initial synchronous sampling dataset serves as the input for subsequent multi-channel alignment and noise subtraction processing.

[0057] After obtaining the initial synchronous sampling dataset, multi-channel time alignment, common-mode noise subtraction, and temperature value binding are performed on the dataset to form a stable reference for subsequent operational analysis. Specifically, multi-channel time alignment involves standardizing the time labels of all channels. Based on the channel number registration results in the array deployment record, the data of each channel in the same sampling round are rearranged according to a unified time series order. Sample segments with missing time labels are removed and marked as invalid segments, ensuring that each working probe, reference probe, and temperature measurement node has a consistent time axis within the same sampling round. Common-mode noise subtraction involves using the background electric field response signal recorded by the reference probe as a common background component and subtracting this common background component from the corresponding time readings of each working probe without modifying the original amplitude records. This process yields differential components reflecting the local electric field response. Simultaneously, abnormally high-amplitude jump segments appearing after subtraction are anomaly-marked, and the corresponding channel's environmental interference status is recorded in the array deployment record. Temperature value binding processing involves, after completing multi-channel time alignment and common-mode noise subtraction, finding the corresponding temperature measurement node within the vicinity of the coordinate binding area for each working probe or reference probe. The initial temperature sampling result of this temperature measurement node is then associated one-to-one with the differential component of the working probe or reference probe in the same sampling round, ensuring that each electric field reading segment carries a corresponding temperature reading. The processing results obtained through multi-channel time alignment, common-mode noise subtraction, and temperature value binding are recorded as a spatial electric field-temperature joint baseline model. This model serves as a reference structure describing the correspondence between the spatial electric field distribution and temperature distribution around the target battery module under nominal static or low-power stable operating conditions. It includes the differential component range, background noise level, coordinate position, and temperature reading range of each working probe and reference probe. The space electric field-temperature joint baseline model is output as an output field name in this step and used as input in subsequent steps. In these subsequent steps, the runtime synchronous acquisition scheduling operation and channel-by-channel deviation comparison processing both use the space electric field-temperature joint baseline model as a reference standard, thereby forming the runtime synchronous acquisition data packet and the list of disturbance candidate events. Through the above processing, the space electric field-temperature joint baseline model constitutes the baseline state description for subsequent steps, guiding the runtime sampling strategy, deviation verification strategy, and region attribution strategy. Therefore, this output field name can be directly referenced in subsequent steps and used as the input location for the next step.

[0058] In one embodiment, specifically in an engineering implementation, for a lithium-ion battery module (numbered BAT_01A03) in row A, column 03 of energy storage compartment 1 of a certain energy storage power station, the system executes step S100 to establish its spatial electric field-temperature joint baseline model. First, maintenance personnel install a sensor array around the module according to a predefined scheme: working probes P_W_01 and P_W_02 are fixed 5cm above the positive and negative terminals of the module housing, respectively; working probe P_W_03 is installed at the center seam on the side of the housing; and a reference probe P_R_01 is installed on the compartment support 50cm away from the module. Simultaneously, a PT100 temperature sensor is attached to the center of the top surface of the module housing, the positive terminal side housing, and the negative terminal side housing, respectively numbered T_01A03_Top, T_01A03_Pos, and T_01A03_Neg. After installation, the system performs probe fixed position registration to confirm that all probes are securely installed; performs number registration to bind the physical probes to the acquisition channel addresses (e.g., AI_CH1 corresponds to P_W_01); and performs coordinate binding to establish a local coordinate system with the module's geometric center as the origin. For example, the coordinates of P_W_01 are registered as (x=0mm, y=+45mm, z=+100mm) and bound to the region "positive pole neighborhood".

[0059] Subsequently, under the stationary operating conditions of the power station at night (system SOC=50%, no charging / discharging power), a synchronous sampling test was triggered. The system simultaneously acquired 10 seconds of data from all probes (sampling rate 100kSPS). The background noise was extracted, processed, and recorded to P_R_01, with the root mean square value of the background electric field noise stabilizing at 0.05V / m. Initial temperature sampling showed readings of 25.3°C, 25.8°C, and 25.1°C at the three temperature nodes. Next, the system performed multi-channel time alignment to ensure that the timestamps of all channel data were synchronized; common-mode noise subtraction was performed, for example, subtracting the signal of P_R_01 from the original signal of P_W_01 point by point to obtain its differential signal; temperature value binding was performed, associating the temperature curve of T_01A03_Pos with the differential signal of P_W_01. Finally, the system generated a joint baseline model of the spatial electric field and temperature for module BAT_01A03. The model records that, under healthy, static conditions, the normal fluctuation range of the differential signal of probe P_W_01 is ±0.15V, with a corresponding temperature reference of 25.8°C ±0.5°C. This model, as output, will provide a reference benchmark for signal comparison during the subsequent S200 steps.

[0060] The technical effects of this step can be summarized as follows: By centrally accessing the configuration data of the space electric field sensing array, the configuration data of the temperature measurement node, and the location information of the target battery module, and combining the fixed position registration, number registration, coordinate binding, synchronous sampling test, background noise extraction, initial temperature sampling, multi-channel time alignment, common mode noise subtraction, and temperature value binding processing, a space electric field-temperature joint baseline model is formed, and this space electric field-temperature joint baseline model is used in subsequent steps.

[0061] S200: Extract the working probe signal, reference probe signal and temperature measurement node signal, perform bandwidth-limited filtering and noise reduction, time synchronization in the same round, differential calculation and channel-by-channel deviation comparison processing, and perform common-mode subtraction correction on the background field signal output by the reference probe to obtain a list of disturbance candidate events;

[0062] This step relies on the spatial electric field-temperature joint baseline model output from the previous step S100. This model records the electric field distribution fingerprint, background noise level, temperature measurement node reading range, and probe number and coordinate binding relationship of the target battery module under nominal static or low-power stable operating conditions. During runtime, synchronous acquisition actions are scheduled for the same target battery module to construct a runtime synchronous acquisition data package. This runtime synchronous acquisition data package refers to simultaneously pulling data from all working probes, reference probes, and temperature measurement node channels under charging, discharging, equalization, and standby operating scenarios, archiving the original timing readings, sampling trigger time markers, channel numbers, and coordinate binding information. The synchronous acquisition action is triggered by the acquisition master control logic, which reads the probe number and coordinate binding relationship recorded in the spatial electric field-temperature joint baseline model and wakes up the corresponding reading channel for each probe according to this relationship, forming a batch of sampling segments. If abnormal situations such as channel reading loss, channel saturation, or reading values ​​exceeding the upper limit of the sampling hardware range occur during the sampling process, the corresponding channel reading segment in the synchronous acquisition data packet during operation will be marked as an invalid segment, and the channel number and time label of the invalid segment will be recorded. This will be used as an anomaly masking reference in the subsequent differential calculation stage to prevent the invalid segment from being regarded as the source of real local field strength jumps in the subsequent differential calculation stage.

[0063] After constructing the synchronous data acquisition data during operation, the working probe signal, reference probe signal, and temperature measurement node signal are extracted from the data acquisition data. Then, filtering and denoising processing, time synchronization processing, and differential calculation processing are performed sequentially to construct a local differential electric field disturbance signal sequence. The working probe signal is a non-contact electric field reading from a local high-risk area, such as the high potential difference region near the target battery module casing, the casing seam region, or the region near the electrode post. The reference probe signal is a background electric field reading located far from the high potential difference region and the strong coupling region. The temperature measurement node signal is a temperature reading close to the outer surface of the casing or adjacent to the inner wall of the casing. In the filtering and denoising processing, the original signals of each channel in the synchronous data acquisition data are cleaned through a signal cleaning process within a preset bandwidth range to remove power frequency coupling interference, communication interference, and broadband burst high-amplitude noise. In the time synchronization processing, for the time tags of different channels, the main control logic reconstructs the corresponding reading segments of all channels in the same sampling round to the same standard time axis according to the numbering registration rules and channel sampling order records in the spatial electric field-temperature joint baseline model. If there are missing segments, a time gap marker is registered at the segment position without interpolation. In the differential calculation process, the reference probe signal is considered as a common background field component. The portion of the working probe signal that is in phase with this common background field component is subtracted, retaining only the differential component that can express the local anomalous electric field distribution. If a working probe still exhibits an abnormally steep pulse segment after this round of differential calculation, the segment is temporarily registered as a local high-sensitivity disturbance segment. After the above filtering and denoising processing, time synchronization processing, and differential calculation processing are executed in sequence, the resulting local differential electric field disturbance signal sequence consists of multiple calibrated working probe differential components. Each differential component carries a time label, coordinate binding information, and a corresponding temperature measurement node number. The local differential electric field disturbance signal sequence is thus archived as an intermediate product for subsequent channel-by-channel deviation comparison processing module in this step.

[0064] After obtaining the local differential electric field disturbance signal sequence, a channel-by-channel deviation comparison process is performed. This process involves retrieving the corresponding space electric field-temperature joint baseline model record for each working probe in the current operating cycle, comparing the amplitude shape, duration, and preceding and following envelope trends of the current differential component with the normal differential range archived in that record. If the current differential component exhibits sudden spikes, sustained spike intervals, or stepped rise segments within the same coordinate binding region, and these features are not within the normal differential range registered in the space electric field-temperature joint baseline model, then that time period is identified as a suspected discharge disturbance segment. These suspected discharge disturbance segments are packaged into event entries, each containing the probe number, coordinate binding region, time label, differential component segment, and corresponding temperature measurement node number. During runtime, the channel-by-channel deviation comparison process simultaneously checks the temperature measurement node signal readings near the specified time marker. If the temperature measurement node signal exhibits a sudden spike or a slow upward trend, a temperature indicator is added to the event entry, indicating that the suspected discharge disturbance segment is accompanied by localized heat accumulation characteristics. These heat accumulation characteristics reflect the heating status of the structure near the insulation degradation segment. If the reference probe signal shows a consistent slow drift across all channels, while the working probe signal, after differential analysis, does not have an independent spike, this round of readings is marked as a global background disturbance segment and is not written into the event entry. If a working probe exhibits continuous saturation output, this channel is marked as an invalid channel in this round of channel-by-channel deviation comparison. This invalid channel record is stored in the extended record area of ​​the synchronously acquired data packets during runtime to prevent subsequent steps from mistaking hardware saturation behavior for a genuine discharge disturbance.

[0065] Through the aforementioned channel-by-channel deviation comparison processing, all suspected discharge disturbance segment event entries are merged according to the time label order and coordinate binding region order to form a disturbance candidate event list. The disturbance candidate event list is the final output field name of this step. This list lists suspected discharge disturbance segments one by one, and binds each suspected discharge disturbance segment with a probe number, coordinate binding region, associated temperature measurement node number, recording time label, and differential component segment index. The disturbance candidate event list can be directly retrieved by downstream calling modules. The operating logic is as follows: subsequent step S300 reads duplicate event entries within the same coordinate binding region from the disturbance candidate event list, performs screening for repeated occurrences at the same location, amplitude evolution statistics, and persistence marking processing on the duplicate event entries, thereby generating time feature annotation results; this sequential relationship completes the cross-main step connection from S200 to S300. The disturbance candidate event list also maintains an index association with the runtime synchronous acquisition data packets. The runtime synchronous acquisition data packets are still used as input in subsequent step S300 to restore the original channel timing, temperature measurement node signals, and other supplementary records, avoiding the loss of contextual information.

[0066] In one embodiment, during constant current charging in an afternoon, the system performs real-time monitoring of module BAT_01A03 in step S200. A synchronous data packet was collected during operation, recording 30 seconds of synchronization data. The system extracts signals from this data packet: first, bandwidth-limited filtering and noise reduction are performed on the working probe signals P_W_01, P_W_02, and P_W_03, the reference probe signal P_R_01, and all temperature node signals, using a bandpass filter with a cutoff frequency of 1kHz-300kHz to suppress charger switching noise (>500kHz) and power frequency interference (50Hz).

[0067] After filtering, time synchronization and differential calculations are performed for the same round. For example, the differential signal V_diff_02(t) of P_W_02 (negative electrode probe) is calculated as V_diff_02(t) = Raw_P_W_02(t) - Raw_P_R_01(t). Subsequently, channel-by-channel deviation comparison processing is performed: V_diff_02(t) is compared with the normal range (±0.15V) of the probe recorded in the baseline model generated by S100. The system found that within a short period of 11 milliseconds from timestamp T=15.234s to T=15.245s, V_diff_02(t) exhibited a rapidly rising and falling pulse, with a peak value reaching +0.92V, far exceeding the normal threshold. At the same time, the associated temperature node T_01A03_Neg signal was checked, and no significant temperature rise (25.2°C) was found during this period. The system determined that this was a local electric field disturbance, not caused by an overall temperature rise. Therefore, a disturbance candidate event entry is generated, containing [Probe number: P_W_02, Time tag: 15.234s-15.245s, Coordinate region: Negative pole neighborhood, Peak amplitude: +0.92V, Associated temperature node: T_01A03_Neg, Temperature change flag: None]. In this round of monitoring, three similar brief pulse events were identified on module BAT_01A03. The system arranges them in chronological order to generate a disturbance candidate event list. This list, as the output of this step, will be acquired by step S300 for further temporal and spatial feature analysis.

[0068] The technical effect of this step can be summarized as follows: This step relies on the space electric field-temperature joint baseline model to drive the synchronous acquisition action during operation. The synchronous acquisition data packet during operation carries multi-channel readings under the field conditions. After filtering and noise reduction, time synchronization, differential calculation and channel-by-channel deviation comparison, the output field name of the disturbance candidate event list is obtained. This disturbance candidate event list is then handed over to the subsequent step S300 for invocation, thereby transforming the instantaneous spatial electric field abnormal behavior around the battery module into a structured candidate event record and entering the next stage of the time-space interpretation process.

[0069] S300 performs screening for repeated occurrences at the same location, amplitude evolution statistics, and persistence marking, and clusters and registers spatial adjacency relationships for records from adjacent working probes. It also performs stage-based judgment processing of spatial feature annotation results and temperature measurement node signals to generate risk level records.

[0070] Specifically, this step receives the list of candidate disturbance events and the synchronous acquisition data package during operation from the output of the previous step S200 as input. The list of candidate disturbance events is a set of event entries identified as suspected discharge disturbance segments in the channel-by-channel deviation comparison processing. Each event entry records the probe number, corresponding coordinate binding area, time label, differential component segment index, and temperature indication flag. The synchronous acquisition data package during operation is the original time-series sampling content synchronously acquired by all working probes, reference probes, and temperature measurement nodes within the same operation cycle, and retains the time label, sampling order, sampling channel status, and real-time reading information of the temperature measurement nodes. The working probe is a non-contact electric field sensing unit arranged in the high potential difference region, shell seam region, or adjacent region of the electrode post of the target battery module shell. The reference probe is a non-contact electric field sensing unit arranged away from the high coupling region. The temperature measurement node is a temperature sensing unit attached to the outer wall of the target battery module shell or the inner wall of the adjacent cavity to obtain the temperature reading at that location. The target battery module is the battery module unit to be monitored within the electrochemical energy storage system. By obtaining the above input, this step first performs a screening process for repeated occurrences at the same location on the list of candidate disturbance events. This screening involves retrieving the probe number, coordinate binding area, and time label of multiple event entries within the same coordinate binding area, and statistically analyzing whether the same coordinate binding area is repeatedly recorded as a suspected discharge disturbance segment in consecutive operating cycles or multiple time periods within the same operating cycle. Specifically, the coordinate binding area of ​​each event entry in the candidate disturbance event list is used as the region primary key, and the time labels are arranged chronologically to generate a time series divided by region. Under each region primary key, entries with adjacent or similar time labels are merged to determine if there is a trend of repeated triggering in that region. If so, a persistence marker is generated for that region. The persistence marker is a segmented identifier of the repeated triggering behavior in that region, and the number of recurrences, the interval between adjacent triggers, and the span of the trigger distribution are written in the persistence marker. During the screening process described above, if the probe number corresponding to an event entry is recorded as an invalid channel or a saturated reading in the synchronous data acquisition data during operation, then this event entry will not be counted in the recurrence count in this round of same-location recurrence screening, and the entry will be retained as a single-occurrence abnormal entry without triggering a persistent marker. Through this same-location recurrence screening process, persistent marker results are obtained for each coordinate-bound region, and these persistent marker results are incorporated into subsequent amplitude evolution statistics operations.

[0071] Further, after obtaining the aforementioned persistent labeling results, this step performs amplitude evolution statistical processing and persistent labeling processing on each event entry in the disturbance candidate event list. Amplitude evolution statistical processing refers to locating multiple differential component segment indices with the same probe number from the disturbance candidate event list, and combining them with the original time-series sampling content in the synchronous acquisition data packet during operation to reconstruct the local differential electric field disturbance amplitude change trajectory at the corresponding time. Specifically, for each event entry, the differential component segment index associated with that entry is extracted, the differential components of the working probe signal for the corresponding time period in the synchronous acquisition data packet during operation are read to form a continuous amplitude trajectory, and then this amplitude trajectory is spliced ​​with the same type of trajectory with adjacent time labels to obtain a description of the amplitude evolution curve of a certain coordinate-bound region within a certain time span. The results of amplitude evolution statistical processing are structured into time feature annotation results in this step. The time feature annotation results are time axis descriptions output for each coordinate-bound region, recording information such as the number of recurrences of the coordinate-bound region, the peak amplitude level of a single disturbance, the duration of the disturbance, and the time labels of the start and end of the disturbance. Persistence marking is performed based on the time feature annotation results. Coordinate-bound regions exhibiting recurring behavior are identified as persistent, and the persistence marker is written into the region entry of the time feature annotation results to indicate whether the region continuously experiences local differential electric field perturbations. If a coordinate-bound region only experiences a single high-amplitude spike without subsequent records, the region is marked as a single high-amplitude event in the time feature annotation results; if the same coordinate-bound region repeatedly experiences medium-amplitude perturbations under different time labels, the region is marked as a periodic recurring event in the time feature annotation results. Understandably, the above time feature annotation results are retained as intermediate products. Since persistence markers have already been aggregated in the time feature annotation results, the time feature annotation results no longer need to directly reference all original differential component fragment indices and all original time-series sampling content. Instead, they reference their summary descriptions and indicate the corresponding coordinate-bound regions and their temporal behavior patterns.

[0072] Subsequently, after obtaining the temporal feature annotation results, this step performs clustering and spatial adjacency registration processing on adjacent working probe records to generate spatial feature annotation results. Adjacent working probe records are defined in the array deployment record as groups of multiple working probes located in spatially proximal areas and sharing the same battery casing surface or the same casing edge segment. Spatial adjacency registration processing refers to determining whether two or more working probes are located on the same casing surface, at the same casing angle, in the same pole neighbor, or in a continuous area on the casing surface, based on the coordinate binding relationships in the array deployment record. Specifically, this step matches each coordinate-bound area in the temporal feature annotation results with the working probe numbers within that area in the array deployment record to determine the set of working probes belonging to the same spatial adjacency range, forming clusters. Under each cluster, the corresponding casing region coordinates, the set of working probe numbers involved, and their corresponding persistence markers and amplitude evolution trajectory summaries in the temporal feature annotation results are recorded, thereby generating spatial feature annotation results. In this step, the spatial feature annotation result is defined as a summary description of a specific physically adjacent region. This summary description merges the temporal feature annotation results of multiple working probes within the same physically adjacent region and retains the spatial adjacency registration information for that region. This spatial adjacency registration information describes whether the region contains continuous anomalous bands, boundary-type anomalous bands, or isolated point anomalous bands. If a working probe is marked as an invalid channel in the synchronous data acquisition packets during operation, that probe will not be included in the valid clustering grouping within that physically adjacent region. However, it will be recorded as an unavailable node and marked as such in the spatial feature annotation result. This avoids misclassifying the unavailable node as a breakpoint in a low-risk or high-risk region during subsequent stage judgment processing. Understandably, through the above clustering grouping and spatial adjacency registration processing, the spatial feature annotation result integrates the perturbation behavior at the single-point probe level into a planar or strip-shaped regional expression, facilitating stage judgment processing based on regions rather than single points in subsequent steps.

[0073] After obtaining the spatial feature annotation results, this step further performs stage determination processing on the spatial feature annotation results and the temperature measurement node signals in the synchronous data acquisition data during operation. The temperature measurement node signal is the temperature reading sequence of the temperature measurement node recorded in the synchronous data acquisition data during operation. This temperature measurement node is located near the outer wall of the target battery module housing or the inner wall of the cavity near the housing, and is used to collect the local temperature evolution at that location. For each physically adjacent region in the spatial feature annotation results, this step queries the temperature measurement node number corresponding to that physically adjacent region in the array deployment record, and then reads the corresponding temperature reading time series of that temperature measurement node from the synchronous data acquisition data during operation. During the stage determination processing, the temperature reading time series is compared in parallel with the persistence markers and amplitude evolution trajectory summaries in the time feature annotation results. If the spatial feature annotation results indicate that the physically adjacent region has persistent markers accompanied by recurring medium-amplitude perturbations, and the corresponding temperature measurement node signal remains within the range of the initial temperature sampling within the same time window, then the physically adjacent region is identified as an early weakening tendency region in the stage determination processing. If the spatial feature annotation results indicate a significant increase in the amplitude evolution trajectory of the physically adjacent region, with persistent marking showing multiple repetitions, and the corresponding temperature measurement node signal shows a slow rise within the same time window, then the physically adjacent region is identified as an insulation degradation region accompanied by local energy accumulation during the stage determination process. If the spatial feature annotation results indicate that the physically adjacent region experiences a high-amplitude, short-duration spike that quickly disappears, and the corresponding temperature measurement node signal shows a momentary jump after the spike ends, then the physically adjacent region is identified as a suspected precursor region of local breakdown during the stage determination process. The above stage determination process does not introduce new probe numbering rules, but directly reuses the consistent numbering relationship and coordinate binding relationship between the disturbance candidate event list, time feature annotation results, spatial feature annotation results, and temperature measurement node signals, avoiding deviation from the spatial reference system of the target battery module.

[0074] Through the aforementioned stage-based judgment process, the judgment result obtained in this step is recorded as a risk level record, which is the output field name of this step. The risk level record generates risk level entries for specific areas of the target battery module's outer surface or adjacent cavities. Each risk level entry carries area coordinates, cluster grouping identifiers, persistence markers, amplitude evolution trajectory summaries, temperature measurement node signal correlation results, and stage-based judgment conclusions. The risk level record is output as an output field name by this step and is read by subsequent step S400. After receiving the risk level record, subsequent step S400 performs maintenance inspection marking processing based on the area coordinates and stage-based judgment conclusions, load reduction instruction template matching processing, and isolation instruction template matching processing to form a graded disposal draft. Then, it generates a graded disposal instruction set and registers the module-level execution status, and finally writes back the space electric field-temperature joint baseline model to construct a closed loop. In essence, the risk level record completes the cross-main-step inheritance relationship from the disturbance candidate event list in the preceding step S200 to the graded disposal instruction set in the subsequent step S400, and serves as the sole trigger source input to the subsequent disposal process.

[0075] In one embodiment, the system acquires the list of disturbance candidate events for module BAT_01A03 and the corresponding runtime synchronous acquisition data packets generated in step S200, and performs analysis in step S300. First, a screening for recurring events at the same location is performed: the system finds that in the runtime data of the last 2 hours, in the coordinate region bound to the "negative pole neighborhood" (probe P_W_02), in addition to the events mentioned above, four similar microsecond-level pulse events were recorded at earlier time points. Persistent labeling processing summarizes these events, marking the region as having "repeated triggering" characteristics, with a trigger count of 5 and adjacent intervals between 10-30 minutes.

[0076] Next, amplitude evolution statistics were performed: the system traced back the original differential signal waveforms corresponding to these 5 events from the synchronously acquired data packets during operation, and calculated the peak amplitudes of the events to be 0.35V, 0.41V, 0.67V, 0.52V, and 0.92V, respectively, showing an overall upward trend. Clustering and spatial adjacency registration processing revealed that the working probe P_W_03 (at the shell seam), which is adjacent to P_W_02, also recorded two weak disturbances in a similar time period. Therefore, the "negative pole-seam area" where P_W_02 and P_W_03 are located was clustered into a single spatial feature labeling result, indicating an anomaly in spatial correlation in this area.

[0077] Subsequently, the system performs a phased judgment process: It queries the long-term sequence signals of the temperature nodes T_01A03_Neg and T_01A03_Top associated with this clustered region. Data shows that over the past two hours, the temperature in this region has experienced a slow, monotonous increase of approximately 1.5°C, which is inconsistent with the overall ambient temperature rise trend. The system compares the "repetitive and intensifying electric field disturbances" with the "continuous localized temperature rise" in parallel, determining that the region is no longer limited to early transient discharge but has entered a stage of insulation degradation accompanied by localized energy accumulation.

[0078] Finally, the system generates a risk level record containing the following core fields: [Regional coordinates: BAT_01A03 negative pole - joint area, cluster grouping identifier: G_01, persistence marker: repeated 5 times / amplitude increase, temperature correlation result: gradual increase of 1.5°C, stage judgment conclusion: insulation degradation area (medium risk)]. This record, as the output of this step, will directly drive the graded treatment in the S400 step.

[0079] The technical effect of this step can be summarized as follows: By screening for repeated occurrences at the same location, statistically analyzing amplitude evolution, marking persistence, registering spatial adjacency relationships, and processing stage judgments in the list of disturbance candidate events and the data packets collected synchronously during operation, a risk level record is generated as an output field name. This translates the local differential electric field disturbance behavior around the target battery module into a clear regional risk level entry, which can be directly called by the subsequent step S400 to carry out graded disposal.

[0080] In one embodiment, specifically, this step first takes the list of disturbance candidate events output from the preceding step S200 and the synchronous data acquisition data during operation as input. The list of disturbance candidate events records the probe number, coordinate binding area, time label, differential component segment index, and temperature indication flag for each item. The synchronous data acquisition data during operation contains the original time-series sampling content and channel status of all working probes, reference probes, and temperature measurement nodes within the same operation cycle. To establish a repetitive triggering metric, a binary event sequence sorted by time label is constructed for each coordinate binding area, and the sequence is convolved and statistically analyzed within a fixed time window to form a verifiable intermediate index of repetition at the same position. The convolution kernel width and time window length are both given by the sampling beat and upper-level strategy table in the aforementioned synchronous data acquisition data during operation. Subsequently, for the items that pass the screening, their corresponding differential component segments are traced back from the synchronous data acquisition data during operation to reconstruct the amplitude trajectory and record the disturbance duration, peak value, and start and end time labels. Based on these time elements and repetition metrics, the time feature annotation results are output as input for the subsequent spatial clustering and stage determination in this step. To make the screening quantifiable and traceable, this step uses convolution and cross-correlation in time-series signal tools to aggregate binary events and suppress occasional noise. The mathematical relationship is expressed by the following formula. Formula (1) is for the binary sequence of events in each coordinate-bound region. With length Normalized window The repetition intensity is given by formula (1).

[0081] ,

[0082] in:

[0083] Indicates the superscript index of the coordinate binding region;

[0084] For continuous time;

[0085] For evaluation time;

[0086] The trajectory is a binary event.

[0087] For window length Internally normalized rectangular or Hanning window functions;

[0088] The length of the time window;

[0089] This indicates the intensity of repetition at the same location.

[0090] Data source → Metrics → Variable mapping:

[0091] The time-labeled sequence extracted from the list of perturbation candidate events is denoted as follows: The time window parameters extracted from the threshold and strategy table are denoted as follows: and Together they form formula (1) Formula (1) yields As an intermediate measure for screening recurrence at the same location in S300, it is directly called in subsequent formula (2). In order to combine the consistency of recurrence intensity and trigger interval, this step calculates the average interval of adjacent events in the same area and performs weighted fusion to obtain the persistence score: formula (2)

[0092] ,

[0093] in:

[0094] The weighted coefficients are derived from the threshold and the policy table, satisfying... ;

[0095] This is the average of adjacent trigger intervals in the same region, calculated from the list of disturbance candidate events → time stamps;

[0096] This is a smoothing constant for intervals;

[0097] For continuous scoring.

[0098] Data source → Metrics → Variable mapping:

[0099] The trigger interval set is obtained by extracting the list of disturbance candidate events. Extracted from threshold and strategy table And the formula (1) obtained Substitute the first term of formula (2) and output the result. The output of this section is the time feature annotation result, which includes... and The time series and associated peak, duration and start and end time labels; the time feature annotation result is recorded as an intermediate product of this step and is called by S300 to perform clustering and spatial adjacency registration of records from adjacent working probes, which is equivalent to the input item of the spatial feature annotation result of S300.

[0100] Following the aforementioned temporal feature annotation results, this section performs clustering and spatial adjacency registration on the records of adjacent working probes, and conducts message passing on the spatial adjacency graph to obtain a persistent measure of neighborhood consistency. Specifically, based on the coordinate binding relationships in the array deployment records, adjacency weights are constructed between the regions to which the working probes belong. (Based on array deployment coordinates and distance thresholds), positive weights are assigned only when two regions are on the same shell surface or share a shell edge segment; for each region, the temporal feature annotation results are used to... As initial values, a normalized aggregation is performed on the adjacency graph to obtain a persistence index for spatial smoothness. This process is equivalent to adjacency normalization and one-time message passing in graph tools, used to suppress occasional disturbances of isolated points without losing region boundary information. To quantify aggregation relationships, the following formula is given: Formula (3)

[0101] ,

[0102] in:

[0103] The adjacency weights are generated by the array deployment records → inter-region distances and topological relationships, and are set to zero for non-adjacent regions;

[0104] For regional indexes;

[0105] It serves as a persistence indicator after spatial aggregation.

[0106] Data source → Metrics → Variable mapping:

[0107] Extracted from time feature annotation results And map to the corresponding region The adjacency relationship is extracted from the array deployment record to generate the array. The formula obtained from formula (2) Substitute directly into the numerator of formula (3). To characterize banded anomalies and boundary anomalies, a spatial consistency score including gradient penalty is further constructed. The ability to express non-uniform regions is enhanced by neighborhood difference terms: Formula (4)

[0108] ,

[0109] in:

[0110] As an indicator for enhancing spatial consistency;

[0111] The spatial difference weights are derived from the threshold and the policy table;

[0112] For the region and adjacent areas Spatial variability indicates the degree of dispersion in the persistence of two regions;

[0113] This is the function for taking the absolute value.

[0114] Data source → Metrics → Variable mapping:

[0115] Extracted from threshold and policy table The formula obtained from formula (3) and Substituting into formula (4), we get The output of this section is spatial feature annotation results, which include the member set of each cluster group, shell region coordinates, adjacency registration information, and... The regional-level curve; the spatial feature annotation result is recorded as an intermediate product of this step and is used as input by the stage judgment process of S300.

[0116] After obtaining the spatial feature annotation results, this section performs a decision-making process. By fusing the temporal structure of the temperature measurement node signals, a risk level record that can directly drive subsequent actions is generated. Specifically, this involves retrieving data from synchronously acquired data packets during operation that are relevant to the regional... The temperature measurement node signals with consistent numbering are used to calculate the rate of temperature change and the magnitude of temperature deviation within a time window, which are then used as quantifications of heat accumulation; subsequently, the... A joint stage index is formed by linearly weighting the heat accumulation and then mapped to a stage label set via segmented thresholds. Considering the strong temporal nature of the monitoring scenario, this section uses the Continuous Wavelet Transform (CWT) tool to converge energy on the multi-scale changes of the temperature sequence, enhancing the distinction between two different thermal behaviors: slow rise and instantaneous jump. This energy index is then combined with... Linear coupling forms a joint index, as shown in Formula (5).

[0117]

[0118] in:

[0119] The fusion weights from the threshold and policy table satisfy the following conditions: ;

[0120] This is a standardized quantity of multi-scale energy or temperature change rate extracted by CWT from data packets synchronously acquired during operation → temperature measurement node signals.

[0121] This is a joint phase index.

[0122] Data source → Metrics → Variable mapping:

[0123] The result obtained by substituting the spatial feature annotation results into formula (4) Temperature sequences are extracted from data packets collected synchronously during operation and obtained via CWT. Together they form .

[0124] To map the joint stage index to the discrete stage set, a piecewise threshold and a step function (Heaviside) are used to construct discrete stage labels: Formula (6)

[0125] ,

[0126] in:

[0127] For stage labels;

[0128] It is a step function;

[0129] The stage thresholds derived from the threshold and policy table satisfy the following conditions: ;

[0130] Data source → Metrics → Variable mapping:

[0131] Extracted from threshold and policy table And the formula (5) obtained Substituting into formula (6), we get .

[0132] After mapping is completed, each coordinate-bound region is generated with the following information: region coordinates, cluster group identifier, persistence marker, amplitude evolution trajectory summary, temperature measurement node signal correlation results, and stage judgment conclusion. The record entries are merged to form a risk level record; the risk level record is the output field name of this section and is read by the subsequent step S400. It is used for the operation and maintenance inspection mark, load reduction instruction template matching and isolation instruction template matching according to the regional coordinates, and its stage judgment conclusion field is directly used as the entry point for the construction of the S400 graded disposal draft.

[0133] This section summarizes the technical effects: By fusing time-repeated measurements, spatial neighborhood consistency, and temperature multi-scale behavior into a unified link and discretizing them into stage labels, the output risk level records are verifiable and replayable, and maintain a one-to-one correspondence between the numbering and coordinate binding relationship with the preceding and subsequent data structures.

[0134] S400 performs maintenance and inspection marking by regional coordinates, load reduction instruction template matching and isolation instruction template matching, and encapsulates the upper-level management unit output format of the graded disposal draft, performs module-level execution status registration, return status registration and space electric field-temperature joint baseline model update processing, and generates graded disposal instruction set and updated space electric field-temperature joint baseline model.

[0135] This step receives the risk level record output from the preceding step S300 as input. The risk level record is a set of risk level entries generated for multiple regions on the outer surface of the target battery module's casing or the inner wall of the adjacent cavity. Each risk level entry includes region coordinates, cluster grouping identifier, persistence marker, amplitude evolution trajectory summary, temperature measurement node signal correlation results for the corresponding temperature measurement node, and stage judgment conclusion. The stage judgment conclusion is a region-level classification description made during the stage judgment processing of S300 for regions with early weakening tendencies, insulation degradation regions accompanied by local energy accumulation, or suspected local breakdown precursor regions. The risk level record is used to archive and describe potential abnormal insulation discharge behavior region by region without disassembling the target battery module. Specifically, the risk level record is input into the instruction generation process of this step, and firstly, maintenance and inspection marking processing is performed according to the region coordinates contained in the risk level record. The so-called regional coordinate-based maintenance inspection marking process refers to mapping the target battery module casing location or adjacent cavity inner wall location corresponding to each regional coordinate in the risk level record to a maintenance-readable positioning tag. This positioning tag records the regional coordinates, cluster grouping identifier, persistence marker, and stage judgment conclusion, and includes inspection marking information. The inspection marking information indicates that the area needs to be visually inspected, contact safety checked, or casing seal checked by on-site personnel during the next routine maintenance. If the stage judgment conclusion in the risk level record belongs to an area with early weakening tendency, the inspection marking information is written as a regular inspection item; if the stage judgment conclusion belongs to an insulation degradation area accompanied by local energy accumulation, the inspection marking information is written as a priority inspection item; if the stage judgment conclusion belongs to a suspected area of ​​local breakdown precursor, the inspection marking information is written as an immediate inspection item. The above-mentioned regional coordinate-based maintenance inspection marking process forms regional inspection marking items. These regional inspection marking items are still indexed by regional coordinates and inherit the cluster grouping identifier and persistence marker to prevent the loss of spatial adjacency relationships during subsequent scheduling. If the working probe corresponding to a certain risk level item is marked as an invalid channel in the synchronous data acquisition data during operation, an indication that the hardware status needs to be verified on-site will be added to the area inspection mark item. This is used to prioritize confirming the installation status or contact status of the working probe during the actual inspection, so as to avoid the subsequent isolation action being triggered erroneously due to sensor failure.

[0136] Further, after completing the operation and maintenance inspection marking process based on regional coordinates, this step performs load reduction command template matching and isolation command template matching processes on the risk level records to generate a graded disposal draft. The load reduction command template matching process refers to selecting a power reduction strategy, charge / discharge current limiting strategy, or equalization current reduction strategy corresponding to the regional coordinates from a preset load reduction command template library for entries marked as insulation degradation areas accompanied by localized energy accumulation. The load reduction command template library is a set of command paradigms compatible with the upper-level management unit, which is the management device responsible for energy scheduling and power distribution control, typically the centralized control node where the energy storage scheduling controller or Battery Management Unit (BMU) is located. This load reduction command template matching process maps the regional coordinates to the specific battery cell clusters or parallel branches within the target battery module and, combined with persistent markings, determines the duration description or trigger priority description of the load reduction command's effectiveness. The isolation command template matching process calls the isolation command template library for risk level entries marked as suspected precursor areas of local breakdown. This library is a set of templates for module-level isolation strategies, including disconnecting the corresponding branch of the target battery module, suspending the target battery module from power distribution, marking the target battery module as disabled, and recording the disabled time stamp. The process compares each entry in the isolation command template library with the stage-based assessment conclusion of suspected precursor areas of local breakdown, generating isolation command candidates for module-level isolation control. It then binds region coordinates and cluster grouping identifiers to these candidate isolation commands to prevent isolation actions from being triggered on the wrong module or in the wrong physical adjacency area. Understandably, after the load reduction command template matching and isolation command template matching processes are executed, the operation and maintenance inspection marker entries, load reduction command candidates, and isolation command candidates based on regional coordinates are aggregated to form a tiered handling draft. The tiered handling draft consists of multiple entries, corresponding to three handling directions: routine inspection, rate limiting and load reduction, or module-level isolation, and carries fields such as regional coordinates, cluster grouping identifier, persistence marker, and stage judgment conclusion. If there are entries in the risk level record that only belong to areas with early weakening tendencies and do not show a persistence marker, then these entries will only enter the operation and maintenance inspection marker queue in the tiered handling draft and will not enter the load reduction command candidate or isolation command candidate, in order to avoid directly triggering power scheduling intervention or module-level isolation intervention for such areas.

[0137] After the aforementioned draft of the tiered handling procedure is generated, this step encapsulates the draft in the output format of the upper-level management unit, forming a set of tiered handling instructions that can be issued. The upper-level management unit output format encapsulation process refers to converting the regional inspection marker entries, load reduction instruction candidates, and isolation instruction candidates into a message structure that the upper-level management unit can parse. Specifically, firstly, for the regional inspection marker entries corresponding to routine inspections, priority inspections, and immediate inspections, maintenance inspection marker sub-instructions are generated. These sub-instructions contain regional coordinates, cluster grouping identifiers, persistence markers, stage judgment conclusions, and corresponding inspection priority descriptions. After output format encapsulation, they become instruction data that the upper-level management unit can forward to the maintenance work order system. Secondly, for the load reduction instruction candidates, load reduction instruction sub-instructions are generated. These sub-instructions contain the module identifier of the target battery module, a description of the proposed power reduction range, a description of the current limiting for charging and discharging, and a description of the execution duration. They are used to guide the upper-level management unit to implement current limiting and load reduction on the corresponding branch of the target battery module. Next, for the isolation command candidates, isolation command sub-commands are generated. These sub-commands include the region coordinates of the module-level isolation target, cluster grouping identifiers, a description of the stage judgment conclusion as a suspected local breakdown precursor region, a deactivation flag, and a deactivation time stamp. These are used to guide the upper-level management unit to perform module-level isolation control or power allocation removal operations on the target battery module. After the above maintenance inspection flag sub-commands, load reduction command sub-commands, and isolation command sub-commands are merged, a hierarchical handling command set is formed. This hierarchical handling command set is one of the subsequent output field names in this step. The hierarchical handling command set can be directly scheduled and executed by the upper-level management unit, thereby linking the diagnostic results of battery insulation status monitoring to the execution paths of power scheduling, inspection tasks, and isolation actions.

[0138] Furthermore, after the tiered handling instruction set is formed, this step also performs module-level execution status registration and feedback status registration for the tiered handling instruction set. Module-level execution status registration refers to recording whether the target battery module has entered the power reduction state or isolation state after the upper-level management unit issues a load reduction instruction sub-instruction or isolation instruction sub-instruction, and writing the execution start time tag and execution status description. Feedback status registration refers to receiving feedback information from the upper-level management unit or field operation and maintenance terminal. The feedback information includes whether the inspection is in place, whether there is visible damage to the casing, whether there are carbonization marks on the sealing surface, whether the operating temperature response of the target battery module is stable after load reduction, and whether the target battery module has been removed from the power allocation queue after module-level isolation. The above module-level execution status registration and feedback status registration are maintained by this step into an execution closed-loop record structure. The execution closed-loop record structure generates an execution status entry for each area coordinate and cluster grouping identifier, and writes the triggered operation and maintenance inspection mark, the execution status of the load reduction instruction or isolation instruction, and the feedback status. Understandably, this execution closed-loop recording structure is not only a runtime log, but will also be used to update the space electric field-temperature joint baseline model, thereby forming an updated space electric field-temperature joint baseline model. In the preceding step S100, the space electric field-temperature joint baseline model is defined as the initial output field name, representing the reference structure for the space electric field and temperature distribution of the target battery module under nominal static or low-power stable operating conditions. In this step, based on the execution closed-loop recording structure, the space electric field-temperature joint baseline model is updated by taking the state after intervention by the load reduction command sub-command or isolation command sub-command as the new reference state, and re-labeling the background noise level, differential component range, and temperature reading range of the corresponding region coordinates under this reference state. If the clustering group identifier corresponding to a certain region coordinate has been removed from the power allocation queue after the execution of the isolation instruction sub-instruction, then the region coordinate is registered as an isolated region in the updated spatial electric field-temperature joint baseline model, and its original differential component range is marked as a historical range and no longer participates in subsequent normal operating condition comparison; if a certain region coordinate only triggers the load reduction instruction sub-instruction and is not isolated, then the region coordinate is registered as a load-limited region in the updated spatial electric field-temperature joint baseline model, and its differential component range and temperature reading range are used as the new normal range or controlled range for the channel-by-channel deviation comparison processing of the synchronous acquisition data packets in the next round of operation. Through the above update processing, the updated spatial electric field-temperature joint baseline model is another output field name of this step. The updated spatial electric field-temperature joint baseline model is returned to the acquisition control flow of the previous main step and is called by S200 in the next acquisition round, used as a reference for synchronous acquisition scheduling, filtering and noise reduction, time synchronization and differential calculation during operation, thereby forming a closed loop in the battery insulation state monitoring scheme of this invention.

[0139] In summary, the tiered handling instruction set and the updated space electric field-temperature joint baseline model are the two output field names of this step. The tiered handling instruction set is directly executed by the upper-level management unit and is used to carry out operation and maintenance inspection marking, load reduction instruction implementation, and isolation instruction implementation. The updated space electric field-temperature joint baseline model is transmitted back to the operation-period synchronous acquisition and scheduling logic related to S200 and is used in subsequent rounds to judge the degree of deviation of the local differential electric field disturbance signal sequence, thereby maintaining continuity and traceability between different operation cycles.

[0140] In one embodiment, after receiving the risk level record (determined as "insulation degradation area - medium risk") for module BAT_01A03 issued in step S300, the system immediately executes step S400. First, based on the coordinate information in the record, a priority inspection work order is automatically generated in the power plant's digital operation and maintenance system, titled "Check the insulation status at the joint between the negative terminal and the casing of BAT_01A03", and pushed to the mobile terminal of the relevant operation and maintenance personnel.

[0141] Next, the load reduction instruction template matching process is executed. Based on the "medium risk" conclusion, the system matches a preset load reduction instruction template from the instruction template library: "Reduce the maximum allowable charging current of the parallel branch where the target module is located to 50% of the rated value." The system instantiates this template, fills in the specific module number BAT_01A03, and generates an executable load reduction instruction sub-instruction.

[0142] Then, the system encapsulates all handling items into the output format of the upper-level management unit. The encapsulated hierarchical handling instruction set contains two sub-instructions: 1) The above-mentioned load reduction instruction is sent to the Battery Cluster Management Unit (BCU) for execution via the Modbus TCP protocol; 2) The above-mentioned inspection work order information is synchronized to the power plant operation and maintenance management platform via the REST API interface. After the instruction is issued, the system registers the module-level execution status in real time and records "BCU has confirmed receipt of the load reduction instruction, and the current charging current has been limited".

[0143] After a period of time, the maintenance personnel conducted an on-site inspection and reported the status back: The platform recorded that "the sealant on the negative electrode post of module BAT_01A03 showed slight discoloration, but no arcing marks were observed." Upon receiving this report, the system triggered an update of the space electric field-temperature joint baseline model. Given that physical changes had been confirmed in the area and load reduction had been implemented, the system temporarily widened the normal range of the baseline differential signals of P_W_02 and P_W_03 from the original ±0.15V to ±0.6V, and updated the associated temperature reference value to the currently stable temperature value. This generated an updated space electric field-temperature joint baseline model. This model will be used for subsequent monitoring and comparison of the module, thus realizing a closed-loop monitoring and control link from "detecting anomalies" to "implementing measures" and then to "calibrating the reference."

[0144] The technical effects of this step can be summarized as follows: This step calls the risk level record to perform operation and maintenance inspection marking, load reduction instruction template matching, and isolation instruction template matching on the coordinates of each area, generates a graded disposal draft and encapsulates it into a graded disposal instruction set, registers the execution status and feedback status and updates the space electric field-temperature joint baseline model, and outputs the graded disposal instruction set and the updated space electric field-temperature joint baseline model, providing executable control instructions and update references for subsequent cycles.

Claims

1. A method for monitoring abnormal discharge of battery insulation based on a spatial electric field, characterized in that, include: S100: Acquire the configuration data of the space electric field sensing array, the configuration data of the temperature measurement node, and the location information of the target battery module; perform probe fixed position registration, number registration, and coordinate binding processing to generate an array deployment record containing the number, coordinates, and pointing area attributes of each probe; based on the array deployment record, trigger all probes and temperature measurement nodes to perform synchronous sampling tests, and perform background noise extraction, initial temperature sampling, multi-channel time alignment, common-mode noise subtraction, and temperature value binding processing on the sampling results to generate a space electric field-temperature joint baseline model; The probes in the spatial electric field sensing array include working probes and reference probes; the working probes are non-contact electric field acquisition units arranged near the high potential difference region, electrode region, or shell seam region of the battery casing; the reference probes are non-contact electric field acquisition units arranged away from the high potential difference region and the region with a low thermal gradient; the temperature measurement nodes are temperature acquisition units arranged against the outer surface of the target battery module casing or the inner wall of an adjacent cavity. S200. Based on the space electric field-temperature joint baseline model, schedule synchronous acquisition actions to acquire a synchronous acquisition data packet containing readings from all working probes, reference probes, and temperature measurement nodes at the same time. Extract working probe signals, reference probe signals, and temperature measurement node signals from the synchronous acquisition data packet, and perform filtering, noise reduction, and time synchronization processing sequentially. Using the reference probe signal as a common background field component, subtract the common background field component from the signal of each working probe to obtain differential components reflecting the local abnormal electric field distribution. According to the space electric field-temperature joint baseline model, perform channel-by-channel deviation comparison on the differential components of each working probe, identify signal segments that exceed the normal differential range defined by the baseline model as suspected discharge disturbance segments, and generate a list of disturbance candidate events in the form of event entries. The event entry includes the probe number, coordinate binding area, time label, and differential component fragment index; S300: Obtain the list of disturbance candidate events and runtime synchronous data packets; For the list of disturbance candidate events, screening for repeated occurrences at the same location is performed using the coordinate-bound region as the index. The number of events occurring in the same region within a preset time window, the interval between adjacent triggers, and the distribution span are counted. Based on the statistical results, a persistence marker indicating the duration characteristics is generated. Based on the persistent marker, the differential component fragments of each event entry are traced back from the synchronously acquired data packets during the operation period to reconstruct the trajectory of the change in the amplitude of the local differential electric field disturbance, and generate time feature annotation results including the number of recurrences, the peak level of the disturbance amplitude, and the duration of the disturbance. Based on the coordinate binding relationship in the array deployment record, the working probes that are temporally related and spatially located on the same shell surface, shell angle, or pole neighborhood are clustered and grouped. The shell region and working probe number set, persistent markers, and amplitude evolution trajectory summary corresponding to each group are summarized to generate spatial feature annotation results. The spatial feature annotation results are compared in parallel with the temperature measurement node signals of the corresponding coordinates in the synchronously acquired data packets during operation. Based on the continuous marking and temperature evolution trend, the region is determined to be an early weakening tendency region, an insulation degradation region with local energy accumulation, or a suspected local breakdown precursor region, and a risk level record is generated. The risk level record includes regional coordinates, cluster grouping identifiers, persistence markers, and stage judgment conclusions; S400. Obtain the risk level record and execute graded handling instruction matching: Based on the stage judgment conclusion, generate operation and maintenance inspection marking instructions, load reduction instructions, or isolation instructions for areas with different risk levels respectively; wherein, the load reduction instruction includes a power reduction strategy or a charge and discharge current limiting strategy; the isolation instruction includes a strategy of disconnecting the corresponding branch of the target battery module or marking it as disabled. The operation and maintenance inspection marking instructions, load reduction instructions, or isolation instructions are encapsulated into a hierarchical handling instruction set that can be parsed by the upper-level management unit, and then issued for execution; The system receives and registers the execution status feedback information, updates the space electric field-temperature joint baseline model based on the feedback information, and uses the differential component range and temperature reading range under the intervention state as a new reference benchmark to form a closed-loop monitoring and control link.

Images