Energy storage system insulation fault grading early warning method with dynamic threshold adjustment

Abstract

This invention discloses a method for graded early warning of insulation faults in energy storage systems with dynamic threshold adjustment, belonging to the field of intelligent early warning technology for energy storage safety. The method includes the following steps: establishing a potential floating recording band around each energy storage cluster; continuously collecting the ground potential difference, inter-cluster voltage difference, and insulation resistance change trajectory to generate a potential floating original draft; based on the potential floating original draft, performing insulation baseline stripping processing on each energy storage cluster to remove common potential drift factors from the insulation resistance change trajectory, obtaining a net change curve that only reflects the insulation degradation state of that energy storage cluster. This invention achieves effective separation of potential drift factors through the potential floating recording band and insulation baseline stripping processing, accurately reflecting the insulation degradation state of the energy storage cluster itself; and limits the threshold transmission range through threshold isolation boundaries and threshold inheritance blocking control, avoiding threshold reversal, making the graded early warning of insulation faults more accurate and reliable, and improving monitoring stability and safety.

Description

Technical Field

[0001] This invention relates to the field of intelligent early warning technology for energy storage safety, and specifically to a method for graded early warning of insulation faults in energy storage systems with dynamic threshold adjustment. Background Technology

[0002] Dynamic threshold adjustment-based insulation fault classification and early warning for energy storage systems refers to a safety monitoring method that, during the operation of energy storage devices, does not use fixed alarm thresholds but adaptively adjusts the insulation resistance threshold based on real-time changes in multiple parameters such as system operating status, voltage, current, humidity, and temperature. Through data acquisition and trend analysis, the decay rate and fluctuation amplitude of insulation performance are determined, and the early warning thresholds are corrected in real time. Faults are classified and warned according to the degree of deviation and rate of change. Based on this, combined with fault prediction and health management, a health status assessment model incorporating fault prediction and health management functions is established by analyzing historical data and operating trends. This enables dynamic prediction of insulation performance, lifespan assessment, and health trend tracking, ultimately constructing a closed-loop control system integrating early warning, prediction, and health management.

[0003] The existing technology has the following shortcomings: In existing technologies, during insulation status monitoring of multi-cluster energy storage systems, a unified threshold correlation model is typically used for dynamic threshold calculation. This means that each energy storage cluster shares some reference parameters to achieve overall adaptive adjustment. However, when potential fluctuations occur between energy storage clusters, the existing threshold correlation mechanism is prone to coupling misalignment, leading to a shift in the threshold update basis. If the insulation performance of one energy storage cluster becomes abnormal, its corrected dynamic threshold can be incorrectly transmitted to adjacent clusters. This causes some healthy units to inherit the threshold adjustment results of the abnormal units, resulting in threshold reversal. This manifests as low-risk units being mistakenly identified as having high-level alarms, while units with genuine high risks remain silent and undetected. This problem is particularly insidious in dynamic threshold-based early warning systems, leading to misjudgments of warning levels, false triggering of protection links, and failure of high-risk sections, thus severely weakening the reliability and safety protection capabilities of energy storage system insulation monitoring.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for graded early warning of insulation faults in energy storage systems with dynamic threshold adjustment, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for graded early warning of insulation faults in energy storage systems with dynamic threshold adjustment, comprising the following steps: Establish a potential fluctuation recording band around each energy storage cluster, continuously collect the ground potential difference, inter-cluster voltage difference and insulation resistance change trajectory, generate potential fluctuation original draft, and form a unified reference basis for subsequent processing; Based on the original potential fluctuation data, insulation baseline stripping is performed on each energy storage cluster to remove common potential drift factors from the insulation resistance change trajectory, thus obtaining a net change curve that only reflects the insulation degradation state of the energy storage cluster. A cluster threshold mapping list is generated based on the net change curve. An independent dynamic threshold range is configured for each energy storage cluster. The potential interference effect of adjacent energy storage clusters is recorded in the cluster threshold mapping list to form threshold isolation boundary parameters. Threshold inheritance blocking control is implemented based on threshold isolation boundary parameters, which restricts the cross-energy storage cluster threshold transmission path in the cluster threshold mapping list, allowing only the threshold adjustment after low potential interference assessment to enter the dynamic threshold range of the corresponding energy storage cluster. Based on the cluster threshold mapping list modified by threshold inheritance blocking control, insulation fault classification and early warning judgment are performed on each energy storage cluster under the constraint of potential floating recording band, so that insulation performance abnormalities can be dynamically identified and classified for early warning, avoiding the occurrence of threshold reversal phenomenon.

[0007] Preferably, the step of establishing a potential floating recording band around each energy storage cluster includes: A data acquisition path targeting potential changes is deployed around each energy storage cluster. A multi-point parallel acquisition strategy is adopted to continuously acquire the ground potential difference of the energy storage cluster and generate time-series potential change data. Based on the completion of ground potential difference acquisition, continue to acquire the inter-cluster voltage difference between any two energy storage clusters and perform synchronous calibration to construct a complete inter-cluster voltage difference matrix; While collecting the ground potential difference and inter-cluster voltage difference, the insulation resistance change trajectory of the energy storage cluster is continuously collected and an operating environment label is attached. The three types of data are aligned one by one on the time axis. After completing multi-dimensional data collection and integration, the three types of data are organized in a unified time sequence to form the original potential fluctuation data, and a continuous potential change trajectory map is constructed through data domain classification and spatial coding.

[0008] Preferably, during the process of generating the potential fluctuation original draft, the data collected from each energy storage cluster is archived in a unified timeline order, and the data records are simultaneously marked with the acquisition path identifier, time series label, and spatial location code, so that the potential change trajectory of each energy storage cluster in the potential fluctuation recording band has a complete physical correspondence, ensuring that the potential fluctuation original draft maintains continuity and consistency as a unified reference basis in the subsequent processing stage.

[0009] Preferably, the cross bracing steps for the net change curve are as follows: Based on the constructed potential floating original draft, the insulation resistance change trajectory of the energy storage cluster is extracted, and it is correlated item by item with the ground potential difference data in the potential floating original draft to establish the correspondence between potential change and insulation change. Based on the combined data sequence, the potential environment of the energy storage cluster is segmented and extracted to identify common potential drift factors and establish a potential influence spectrum. A correspondence was established between the group of factors to be stripped and the combined data sequence, and insulation baseline stripping was performed to remove common potential drift factors from the insulation resistance change trajectory, generating a net insulation change curve. The net insulation change curves are sequenced and labeled with status to form a data output structure with potential disturbance background markings.

[0010] Preferably, when performing insulation baseline stripping, the potential drift characteristic segment constructed by the ground potential difference is used as the stripping reference. The potential shift influence of the corresponding time period is removed from the insulation resistance change trajectory item by item, so that the net insulation change curve after stripping only reflects the insulation degradation state of the energy storage cluster itself. The sampling order and data field affiliation are kept consistent in the data after stripping to ensure the continuous correspondence between the net insulation change curve and the original potential fluctuation in the data structure.

[0011] Preferably, the threshold isolation boundary parameter generation steps are as follows: A dynamic response range analysis framework is established around the net insulation change curve corresponding to each energy storage cluster. The net insulation resistance change data is divided into time periods and classified into states to construct a response range that matches the insulation degradation behavior of the energy storage cluster. Based on the obtained dynamic response range, configure a dynamic threshold range suitable for the characteristics of the energy storage cluster itself, and set upper and lower limit ranges with buffering characteristics. While constructing independent dynamic threshold ranges, the potential interference effects of adjacent energy storage clusters are recorded, and the interference information is appended to the threshold range configuration field of the energy storage cluster. The parameter fields corresponding to the energy storage clusters are summarized to form a cluster threshold mapping list, and the threshold isolation boundary parameters are formed by field fusion.

[0012] Preferably, during the formation of the threshold isolation boundary parameters, the cluster threshold mapping list is hierarchically labeled according to the potential interference level between energy storage clusters. Energy storage clusters with weak potential interference are labeled as allowable threshold adjustment areas, while energy storage clusters with strong potential interference are labeled as threshold restriction areas, so as to ensure that the dynamic threshold range of each energy storage cluster maintains an independent applicable range under the potential fluctuation environment.

[0013] Preferably, threshold inheritance blocking control is performed based on threshold isolation boundary parameters to restrict the cross-cluster threshold transfer path in the cluster threshold mapping list, allowing only the threshold adjustment after potential disturbance assessment to enter the dynamic threshold range of the corresponding energy storage cluster. The steps are as follows: The threshold transfer paths between energy storage clusters are identified based on the generated cluster threshold mapping list, and the transfer paths are mapped to threshold isolation boundary parameters to construct a transfer path identification map; Based on the interference impact labels of the transmission path, the threshold adjustment permissions of the path are classified and screened, and threshold blocking control criteria are formulated to limit the threshold transmission range. For the state update requirements of energy storage clusters, a threshold adjustment screening operation based on the interference assessment results is performed, and only the threshold adjustment after potential interference assessment is allowed to enter the dynamic threshold range of the target energy storage cluster. The cluster threshold mapping list is updated around the adjusted dynamic threshold range, and the source of the transmission path, the level of interference impact, and the control status are recorded in the list.

[0014] Preferably, when performing threshold adjustment filtering operations based on interference assessment results, the transmission path of each threshold adjustment is associated with the interference impact label, and an isolation marker is added to the blocked transmission path in the clustered threshold mapping list, so as to continuously restrict the threshold transmission of high interference paths in subsequent running cycles and ensure the independent update and stability of the dynamic threshold range.

[0015] Preferably, based on the cluster threshold mapping list corrected by threshold inheritance blocking control, the steps for determining insulation fault classification and early warning for each energy storage cluster under the constraint of potential floating recording band are as follows: Based on the cluster threshold mapping list corrected by threshold inheritance blocking control, an independent early warning judgment reference set is established for each energy storage cluster, and the dynamic threshold interval is associated with the potential fluctuation recording band. By constraining the potential environment of the energy storage cluster using the potential fluctuation recording band, the net insulation change curve is compared within the potential fluctuation range to determine the characteristics of the state change. Under the constraint of potential floating recording band, the insulation status of energy storage clusters is classified and determined according to the early warning reference set, and the position change of the net insulation change curve in the dynamic threshold interval is mapped to the early warning level. The time series processing and state preservation processing of the graded early warning results of energy storage clusters are carried out to prevent the occurrence of threshold reversal.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention effectively separates the potential drift factor, which is prevalent during the operation of energy storage clusters, from the insulation resistance change trajectory by introducing a potential fluctuation recording band and an insulation baseline separation process. This eliminates interference from overall potential fluctuations in insulation status assessment. By constructing a net change curve that reflects the insulation degradation state of a single energy storage cluster and configuring independent dynamic threshold ranges for each cluster, insulation performance anomalies can be continuously and stably identified under complex operating environments. This significantly reduces false alarms and missed alarms caused by potential coupling, improving the reliability and continuity of insulation monitoring results.

[0017] This invention introduces threshold isolation boundary parameters and threshold inheritance blocking control based on the cluster threshold mapping list, giving the threshold adjustment process clear transmission boundaries and constraints. It only allows threshold changes within the low-potential interference range to affect the corresponding energy storage cluster. By combining potential fluctuation recording bands for graded judgment during the early warning determination stage, it effectively avoids the threshold reversal problem caused by mutual influence between thresholds of different energy storage clusters. This ensures a clear distinction between low-risk and high-risk units in terms of early warning levels, thereby enhancing the overall targeting of early warning results and operational safety assurance capabilities. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the method for graded early warning of insulation faults in energy storage systems based on dynamic threshold adjustment according to the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1 The dynamic threshold adjustment method for graded early warning of insulation faults in energy storage systems, as shown, includes the following steps: Establish a potential fluctuation recording band around each energy storage cluster, continuously collect the ground potential difference, inter-cluster voltage difference and insulation resistance change trajectory, generate potential fluctuation original draft, and form a unified reference basis for subsequent processing; This step revolves around the continuous acquisition and full recording of potential changes. Through structured processing, it ensures that subsequent steps such as insulation baseline stripping, threshold interval allocation, and graded early warning have a complete and consistent potential reference framework. The specific implementation steps are as follows: A data acquisition path targeting potential changes is deployed around each energy storage cluster, employing a multi-point parallel acquisition strategy to continuously record the ground potential difference of each cluster. This sub-step generates continuous time-series potential change data by periodically acquiring the instantaneous potential difference between the energy storage cluster and its ground reference point. During acquisition, it is crucial to ensure that the sampling interval remains consistent with the storage structure to avoid gaps or abrupt changes in the potential fluctuation trajectory due to data loss. Furthermore, in this step, the ground potential difference of each energy storage cluster must be individually labeled and archived, ensuring a clear physical correspondence between the potential changes of each cluster and the entire energy storage cluster. This guarantees a precise data attribution basis for the subsequent formation of the potential fluctuation recording band.

[0022] Building upon the acquisition of ground potential differences for each energy storage cluster, structured recording of inter-cluster voltage differences is then conducted. This sub-step involves bidirectional acquisition and comparison of the instantaneous voltage difference between any two energy storage clusters. The acquisition process must cover any combination of voltage difference pairs between energy storage clusters to form a complete inter-cluster voltage difference matrix. During data organization, corresponding acquisition timestamps and spatial identifiers should be added to the voltage difference data between each group of energy storage clusters, and the data should be synchronously calibrated with the ground potential difference acquisition results to ensure consistency of data across the potential fluctuation recording band on the time axis. The implementation of this sub-step provides essential data support for subsequent extraction of potential coupling relationships within the system and further enriches the data hierarchy of the original draft.

[0023] While collecting the ground potential difference and inter-cluster voltage difference, the insulation resistance variation trajectory of each energy storage cluster is simultaneously collected and categorized. This sub-step requires periodic sampling of the insulation resistance of each energy storage cluster, forming a time-series record consistent with the previous two data structures. The recorded insulation resistance values ​​should include continuous variation data during stable operation, voltage variation, and environmental disturbance phases, with additional temperature and humidity labels for the operating environment to further supplement the potential variation background. During data merging, the insulation resistance variation trajectory should be aligned line by line with the corresponding ground potential difference and inter-cluster voltage difference data, using timestamps as anchors, and a unified data recording format should be constructed to ensure consistency in the logical structure of the three types of data. The execution of this sub-step marks the formal completion of the core data layer of the potential fluctuation recording band, laying the foundation for the formation of the final draft.

[0024] After completing the multi-dimensional collection and integration of ground potential difference, inter-cluster voltage difference, and insulation resistance change trajectories, a unified data integration framework is constructed around these three types of data to form a complete potential fluctuation draft. This sub-step organizes all collected data according to a unified timeline order and constructs a continuous potential change trajectory map, clearly depicting the potential state of each energy storage cluster throughout the recording period. During the draft formation process, all recorded data should be classified into data domains, identified by collection paths, labeled with time series, and encoded with spatial locations to ensure that the dataset maintains a consistent logical structure throughout the process. The potential fluctuation draft not only includes the three basic data types but also the collection order, physical affiliation, and inter-cluster reference relationships of each data field, enabling it to serve as a reference for the potential change state of the entire energy storage cluster and providing a complete, unified, and continuous data foundation for subsequent steps. After the draft is formed, its format should be consistent with the data call interface of each subsequent processing stage to ensure that it continues to function as a unified reference throughout the entire dynamic threshold adjustment process.

[0025] Based on the original potential fluctuation data, insulation baseline stripping is performed on each energy storage cluster to remove common potential drift factors from the insulation resistance change trajectory, thus obtaining a net change curve that only reflects the insulation degradation state of the energy storage cluster. Performing insulation baseline stripping on each energy storage cluster based on the potential fluctuation data is a crucial step in constructing the net value characteristics of cluster insulation status. Through a series of data processing operations based on the potential fluctuation data, common potential drift factors affecting the insulation resistance change trajectory are stripped away, thereby extracting the net change curve that truly reflects the changes in the insulation performance of a single cluster. By comparing the correlation between potential change behavior and insulation resistance evolution trends, the independent degradation characteristics of each energy storage cluster can be effectively separated, preventing the overlap of insulation criteria among multiple energy storage clusters under potential coupling conditions. This lays a stable single-cluster foundation for subsequent dynamic threshold allocation and graded early warning determination. The specific steps are as follows: Based on the constructed potential fluctuation data, the insulation resistance change trajectory of the energy storage cluster during the complete monitoring period is extracted and correlated item by item with the ground potential difference data for the corresponding time period in the potential fluctuation data. During this process, it is necessary to maintain the consistency of the insulation resistance data and the ground potential difference data on the time axis, and to attach the sampling time and cluster number as identifiers to the data record at each moment to establish the correspondence between potential changes and insulation changes. The goal of this sub-step is to construct a trajectory combination containing two layers of information for each energy storage cluster, so that the change in insulation resistance not only reflects its own internal state, but also adds reference information that evolves synchronously with the surrounding potential environment, preparing structured input data for subsequent stripping operations.

[0026] Based on the combined data sequence, common potential drift factors in the potential environment of the energy storage cluster are extracted in segments. Specifically, by aggregating and analyzing the ground potential difference values ​​over multiple time periods throughout the monitoring cycle, potential change characteristics that repeatedly appear in the energy storage cluster and maintain a consistent trend under different operating conditions are identified. These characteristics usually originate from potential disturbances caused by external environmental voltage fluctuations, changes in grounding structure, or other non-cluster-specific factors, and exhibit synchronous or near-synchronous shift behavior across multiple energy storage clusters. By organizing these common potential drift factors and establishing corresponding potential influence spectra by dividing them into time periods, a clear reference boundary can be provided for excluding non-cluster-specific characteristics from the insulation resistance trajectory. The potential drift characteristic segments formed in this sub-step, together with the original trajectory, constitute the group of factors to be stripped, constructing an extraction channel for the intrinsic characteristics of the energy storage cluster's insulation state.

[0027] A correspondence is established between the factor groups to be stripped and the combined data sequence, and insulation baseline stripping is performed to remove common potential drift factors from the insulation resistance change trajectory, eliminating the superposition effect of external potential disturbances on the insulation change performance. In this sub-step, the potential drift characteristic segment constructed based on the ground potential difference is used as the stripping reference, and the potential offset influence within the corresponding time period is removed item by item from the insulation resistance trajectory, thereby generating a net insulation change curve with common offset factors removed. During the stripping operation, it is still necessary to maintain the consistency of data sampling order and data field attribution to ensure that the stripping operation is accurately corresponding at each time point and to avoid introducing misalignment in data structure. Through this sub-step, the original trajectory, which originally mixed system-level potential disturbances and intra-cluster insulation changes, is effectively purified, retaining only the intrinsic change process related to the evolution of the insulation performance of the energy storage cluster itself, providing a clear boundary for setting the clustering threshold in subsequent steps.

[0028] After obtaining the net insulation change curve of the energy storage cluster, its sequence is organized and its status is labeled to construct a data output format for subsequent processing. In this sub-step, the net insulation change data after the stripping operation is completed is rearranged according to the sampling time order, and a potential disturbance background mark is added to each data point to indicate whether there is a strong interference area or whether it belongs to the stable window after potential drift stripping. At the same time, the comparison index between the original trajectory before stripping and the net change curve after stripping is retained, so that subsequent steps can trace the data source when necessary and establish a mapping relationship between the net change curve and the original potential fluctuation. The net change curve after the organization is completed will serve as the basic data structure that uniquely represents the insulation performance degradation state of the energy storage cluster. It will be called as the core input in the subsequent construction of the cluster threshold mapping list and the execution of graded early warning judgment, thereby ensuring that subsequent operations have sufficient physical independence at the insulation criterion level.

[0029] A cluster threshold mapping list is generated based on the net change curve. An independent dynamic threshold range is configured for each energy storage cluster. The potential interference effect of adjacent energy storage clusters is recorded in the cluster threshold mapping list to form threshold isolation boundary parameters. Generate a cluster threshold mapping list based on the net change curve, configure an independent dynamic threshold range for each energy storage cluster, and record the potential interference effects of adjacent energy storage clusters in the cluster threshold mapping list to form threshold isolation boundary parameters. This is a crucial technical step for achieving accurate identification and independent early warning of the insulation status between energy storage clusters. Based on the obtained net insulation change curve, a dynamic threshold configuration logic for energy storage cluster units is constructed in a structured manner, and inter-cluster potential interference factors are introduced as influence indicators. This enables adaptive segmentation of threshold boundaries for each energy storage cluster under the same operating environment, thereby establishing an operable data foundation for subsequent early warning triggering control. The specific steps are as follows: A dynamic response range analysis framework is established based on the net insulation change curve corresponding to each energy storage cluster. The net insulation resistance change data of the energy storage cluster within a continuous operating cycle is divided into time periods and categorized by state. This step, based on the obtained net change curve data, labels the operating state and divides the data into continuous segments according to the amplitude and fluctuation rate of insulation resistance change, clarifying the insulation behavior characteristics corresponding to each time period. In this process, it is crucial to identify key time periods exhibiting stable changes, continuous degradation, or short-term sharp fluctuations. The minimum values, average values, and fluctuation limits within these time periods are then summarized and organized to construct a response range matching the insulation degradation behavior of the energy storage cluster. Through the above classification and statistical analysis, the numerical characteristics of the insulation state of the energy storage cluster under three stages—normal fluctuation, critical boundary, and risk approach—can be preliminarily clarified, providing a hierarchical basis for configuring independent dynamic threshold ranges.

[0030] Based on the obtained dynamic response range of the energy storage cluster, and further considering long-term evolution trends and stage-specific limits, a dynamic threshold range suitable for the cluster's own characteristics is configured. This step defines three different levels of insulation state boundaries within the defined segmented data framework, corresponding to the normal operation stage, the early warning critical stage, and the fault trend stage, respectively. When configuring the dynamic threshold range, the continuous downward and oscillating segments in the net change curve should be fully considered to identify potential misjudgments or omissions, and upper and lower limit ranges with buffering characteristics should be set accordingly. The dynamic threshold range of each energy storage cluster should be compatible with its intrinsic insulation degradation behavior, avoiding direct reference to data from other energy storage clusters to prevent cross-cluster misinterpretation. Through this sub-step, each energy storage cluster obtains a set of well-structured, stage-specific, and dynamically evolving insulation threshold ranges, forming the backbone parameter set in the cluster threshold mapping list, laying the foundation for further construction of isolation boundaries.

[0031] While constructing an independent dynamic threshold range, the potential interference effects of neighboring energy storage clusters are extracted and recorded based on the potential interaction between the energy storage cluster and its location, in order to enhance the anti-interference performance of the energy storage cluster's threshold range. This sub-step, based on the inter-cluster potential difference change information recorded in the original potential fluctuation document, filters and archives the potential disturbance events experienced by the energy storage cluster in multiple operating cycles, identifying neighboring energy storage clusters that may interfere with the insulation status judgment of the energy storage cluster. For each type of interference effect, the interference time, interference amplitude, and interference direction should be used as core indicators for characterization, and their corresponding information should be added to the current energy storage cluster's threshold range configuration field. The introduction of this type of interference information makes the dynamic threshold range of the energy storage cluster not only reflect its intrinsic insulation evolution characteristics, but also include the external disturbance background it experiences in the local potential structure. On this basis, the threshold parameters of each energy storage cluster can dynamically respond to changes in the operating status of surrounding energy storage clusters, forming a cluster mapping list with embedded interference tag system based on potential correlation.

[0032] After configuring independent threshold intervals and labeling interference information, all parameter fields corresponding to each energy storage cluster are summarized into a unified-format cluster threshold mapping list. Threshold isolation boundary parameters are then formed within this list through field fusion. This sub-step uses the energy storage cluster number as an index to uniformly encapsulate the net insulation change curve identifier, dynamic threshold interval parameters, interference impact fields, and associated potential labels, and sorts them according to time sequence and spatial location. During the generation of threshold isolation boundary parameters, based on the potential interference levels of neighboring energy storage clusters labeled in the aforementioned interference impact fields, an independent threshold application range and cross-interference boundary are labeled for each energy storage cluster, clarifying which potential change sources can enter the threshold update mechanism and which should be excluded. The final cluster threshold mapping list possesses a complete data structure, including the independent insulation threshold interval for each energy storage cluster and its interference isolation conditions under its potential environment, providing sufficient foundational support for subsequent implementation of threshold inheritance restrictions and graded early warning triggering.

[0033] Threshold inheritance blocking control is implemented based on threshold isolation boundary parameters, which restricts the cross-energy storage cluster threshold transmission path in the cluster threshold mapping list, allowing only the threshold adjustment after low potential interference assessment to enter the dynamic threshold range of the corresponding energy storage cluster. Threshold inheritance blocking control is implemented based on threshold isolation boundary parameters to restrict the threshold transfer paths across energy storage clusters in the clustered threshold mapping list. Only threshold adjustments assessed after low-potential interference are allowed to enter the dynamic threshold range of the corresponding energy storage cluster. This is a crucial step in ensuring the independence of insulation monitoring and the accuracy of dynamic threshold updates for each energy storage cluster. This step, based on the established clustered threshold mapping list and threshold isolation boundary parameters, constrains the threshold transfer paths between energy storage clusters. By controlling the degree of threshold influence under different potential interference levels, potential misleading factors are eliminated, ensuring that threshold adjustments only occur within acceptable interference ranges. The specific steps are as follows: Based on the generated cluster threshold mapping list, potential threshold transfer paths between energy storage clusters are identified, and these paths are mapped one-to-one with threshold isolation boundary parameters to construct a transfer path identification map oriented towards threshold transfer direction. During this process, based on the energy storage cluster numbers and potential interference information of adjacent energy storage clusters recorded in the cluster threshold mapping list, all combinations of energy storage clusters that are physically adjacent, have related data update timing, or mutually influence each other's operating states are extracted and labeled as potential threshold transfer pairs. After identifying possible transfer paths, an interference impact label is attached to each transfer path based on the interference level indicated by the threshold isolation boundary parameters to indicate the degree of potential disturbance that the path may cause to the target energy storage cluster's threshold update during actual operation. Through this sub-step, a structured analysis of all cross-energy storage cluster transfer paths that may affect threshold updates is completed, establishing a basic mapping framework for subsequent transfer access control.

[0034] Based on the identification of transmission paths, the threshold adjustment permissions for each path are tiered and filtered according to its interference impact label, and corresponding blocking control criteria are established. In this sub-step, based on the potential difference fluctuation range between energy storage clusters and the aforementioned tiered interference level standards, acceptable threshold adjustment boundaries are set for each pair of energy storage clusters, and corresponding blocking strategies are defined when these boundaries are exceeded. Transmission paths with low interference impact levels can be marked as allowable adjustment paths, allowing them to influence the dynamic threshold range of the target energy storage cluster under limited conditions. Transmission paths with medium or higher interference impact levels should be marked as restricted adjustment paths or completely blocked paths, and a restriction field for adjustment input should be set in the cluster threshold mapping list. Through this sub-step, a threshold transmission control mechanism can be established between physically adjacent or logically related energy storage clusters, ensuring that the transmission information from high-interference paths does not incorrectly guide the threshold of the target energy storage cluster, thereby maintaining the independence of the dynamic threshold range of each energy storage cluster.

[0035] After defining the permissions for each transmission path, a threshold adjustment filtering operation based on interference assessment results is performed to address the state update needs of the energy storage clusters within the current operating cycle. This sub-step focuses on threshold update requests caused by changes in the state of the energy storage clusters during actual operation. According to the aforementioned transmission path permission definitions, all threshold update information from other energy storage clusters is compared for interference impact and permissions are verified. For threshold adjustment requests originating from low-potential interference paths, only those assessed as having low impact through isolation boundary parameters are included in the target energy storage cluster's dynamic threshold range for numerical updates; while adjustment requests originating from high-interference paths are directly filtered by the system or redirected as local adjustment feedback. During this step, path tracking information for threshold adjustment records should be continuously maintained to ensure that each threshold change operation has a traceable source label. Through this filtering and execution mechanism, when faced with multi-source threshold adjustment requests, each energy storage cluster only adopts update data carried by paths with controllable interference impact, thereby further enhancing the local effectiveness and accuracy of the control range of dynamic threshold updates.

[0036] Based on the completion of threshold screening and control execution, the cluster threshold mapping list for the current operating cycle is updated around the adjusted dynamic threshold range, and the newly formed threshold transmission structure is embedded into the list record to form a closed-loop tracking chain. This sub-step records the final effective threshold adjustment result for each energy storage cluster, associates it with the transmission path source, interference impact level, and adjustment authority identifier, and forms an updated threshold mapping list with transmission path annotation function. During the update process, isolation markers are added to all transmission paths that are blocked to quickly identify potentially misleading channels in the next cycle or subsequent processing stage, preventing them from mistakenly entering the judgment logic again. The final cluster threshold mapping list not only includes the latest dynamic threshold range and its evolution process for each energy storage cluster, but also clearly records the interference source and control status of all adjustment paths, providing a complete, manageable, and traceable threshold adjustment basis for the entire dynamic early warning process.

[0037] Based on the cluster threshold mapping list modified by threshold inheritance blocking control, insulation fault classification and early warning judgment are performed on each energy storage cluster under the constraint of potential floating recording band, so that insulation performance abnormalities can be dynamically identified and classified for early warning, avoiding the occurrence of threshold reversal phenomenon. Based on the clustered threshold mapping list corrected by threshold inheritance blocking control, insulation fault classification and early warning judgment are performed for each energy storage cluster under the constraint of potential floating recording band. This is a key step in transforming the results of potential decoupling, threshold isolation, and threshold propagation control into actual early warning output. This step uses the clustered threshold mapping list with completed threshold inheritance blocking control as the sole threshold basis, and introduces potential floating recording band as a dynamic constraint background. This ensures that the early warning judgment of each energy storage cluster is always performed under controllable conditions with clear potential correlation, thereby guaranteeing that insulation performance anomalies can be continuously and accurately identified and classified according to level, avoiding the recurrence of threshold reversal phenomenon in the early warning stage. The specific steps are as follows: Based on the clustered threshold mapping list corrected by threshold inheritance blocking control, an independent early warning judgment reference set is established for each energy storage cluster. The dynamic threshold range, threshold isolation boundary information, and acceptable threshold adjustment sources corresponding to that energy storage cluster are centrally organized. In this sub-step, taking the energy storage cluster as the smallest judgment unit, all threshold parameters belonging to that energy storage cluster in the clustered threshold mapping list are extracted separately and associated with the corresponding time period data in the potential fluctuation recording band. This ensures that the insulation state judgment at each moment has a clear threshold reference and potential background description. In this way, it is ensured that threshold information from other energy storage clusters or adjustment results that do not conform to the isolation boundary are not mixed in during subsequent early warning judgment processes, maintaining the singularity and consistency of the judgment basis from the source.

[0038] After constructing the independent early warning reference set, the potential environment of the current energy storage cluster is continuously constrained by the potential fluctuation recording band. The real-time state of the net insulation change curve is compared within the corresponding potential fluctuation range. In this sub-step, the potential fluctuation recording band not only serves as background data but also participates in the judgment process as a constraint condition. That is, at each judgment moment, the data point corresponding to the net insulation change curve is correlated with the potential fluctuation range at the same moment to clarify whether the current insulation state changes under a stable potential background or within a potential fluctuation range. By introducing this constraint relationship, high-level early warnings can be avoided by directly triggering them based on numerical changes during periods of severe potential fluctuations, thereby reducing the risk of misjudgment caused by changes in the potential background and ensuring that the early warning judgment is always synchronized with the actual potential environment.

[0039] Under the constraint of potential fluctuation recording, the insulation status of each energy storage cluster is classified and determined according to an independent early warning reference set. The position change of the net insulation change curve in the dynamic threshold interval is mapped to the corresponding early warning level. In this sub-step, the insulation status needs to be divided into multiple continuous level intervals according to the threshold interval structure pre-configured in the cluster threshold mapping list. Combined with the potential background constraint results obtained in the previous sub-step, the triggering of the early warning level is controlled hierarchically. When the net insulation change curve is in the normal range, only its change trend is recorded without triggering an early warning; when the net insulation change curve enters the early warning critical range and the potential background is in a stable state, a low-level early warning is triggered; when the net insulation change curve further enters the high-risk range and is accompanied by a continuous downward trend, the corresponding high-level early warning is triggered. Through this hierarchical determination method, the insulation performance anomalies of each energy storage cluster can be distinguished according to their degree, rather than being simply judged by a single threshold, thereby improving the hierarchy and interpretability of the early warning output.

[0040] After completing the graded early warning determination for each energy storage cluster, the early warning results are processed through time series analysis and status maintenance to ensure consistency of graded early warning information throughout the continuous operating cycle and to continuously prevent threshold reversal. In this sub-step, the changes in the early warning level of each energy storage cluster, along with the corresponding position of the net insulation change curve, the potential fluctuation recording band interval, and the dynamic threshold interval, need to be recorded synchronously to form a complete early warning evolution trajectory. When the insulation status recovers or fluctuates during subsequent operation, the adjustment of the early warning level still needs to be constrained by the dual constraints of the cluster threshold mapping list and the potential fluctuation recording band to prevent frequent jumps or reversals in the early warning level due to short-term fluctuations. Through this continuous constraint and status maintenance method, it is ensured that low-risk energy storage clusters are not misjudged as high-level early warnings due to abnormal states of other energy storage clusters, while energy storage clusters with genuine insulation degradation risks can be continuously monitored and kept within the corresponding early warning level range, thus eliminating the possibility of threshold reversal during the early warning stage.

[0041] This invention effectively separates the potential drift factor, which is prevalent during the operation of energy storage clusters, from the insulation resistance change trajectory by introducing a potential fluctuation recording band and an insulation baseline separation process. This eliminates interference from overall potential fluctuations in insulation status assessment. By constructing a net change curve that reflects the insulation degradation state of a single energy storage cluster and configuring independent dynamic threshold ranges for each cluster, insulation performance anomalies can be continuously and stably identified under complex operating environments. This significantly reduces false alarms and missed alarms caused by potential coupling, improving the reliability and continuity of insulation monitoring results.

[0042] This invention introduces threshold isolation boundary parameters and threshold inheritance blocking control based on the cluster threshold mapping list, giving the threshold adjustment process clear transmission boundaries and constraints. It only allows threshold changes within the low-potential interference range to affect the corresponding energy storage cluster. By combining potential fluctuation recording bands for graded judgment during the early warning determination stage, it effectively avoids the threshold reversal problem caused by mutual influence between thresholds of different energy storage clusters. This ensures a clear distinction between low-risk and high-risk units in terms of early warning levels, thereby enhancing the overall targeting of early warning results and operational safety assurance capabilities.

[0043] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for grading pre-alarm of insulation fault of energy storage system with dynamic threshold adjustment, characterized in that, Includes the following steps: Establish a potential fluctuation recording band around each energy storage cluster, continuously collect the trajectory of changes in ground potential difference, inter-cluster voltage difference and insulation resistance, and generate the original potential fluctuation record; Based on the original potential fluctuation data, insulation baseline stripping is performed on each energy storage cluster to remove common potential drift factors from the insulation resistance change trajectory, thus obtaining a net change curve that only reflects the insulation degradation state of the energy storage cluster. A cluster threshold mapping list is generated based on the net change curve. An independent dynamic threshold range is configured for each energy storage cluster. The potential interference effect of adjacent energy storage clusters is recorded in the cluster threshold mapping list to form threshold isolation boundary parameters. Threshold inheritance blocking control is implemented based on threshold isolation boundary parameters, which restricts the cross-energy storage cluster threshold transmission path in the cluster threshold mapping list, allowing only the threshold adjustment after low potential interference assessment to enter the dynamic threshold range of the corresponding energy storage cluster. Based on the cluster threshold mapping list modified by threshold inheritance blocking control, insulation fault classification and early warning judgment are performed on each energy storage cluster under the constraint of potential floating recording band, so that insulation performance abnormalities can be dynamically identified and classified for early warning.

2. The method of claim 1, wherein, The steps for establishing potential floating recording bands around each energy storage cluster include: A data acquisition path targeting potential changes is deployed around each energy storage cluster. A multi-point parallel acquisition strategy is adopted to continuously acquire the ground potential difference of the energy storage cluster and generate time-series potential change data. Based on the completion of ground potential difference acquisition, continue to acquire the inter-cluster voltage difference between any two energy storage clusters and perform synchronous calibration to construct a complete inter-cluster voltage difference matrix; While collecting the ground potential difference and inter-cluster voltage difference, the insulation resistance change trajectory of the energy storage cluster is continuously collected and an operating environment label is attached. The three types of data are aligned one by one on the time axis. After completing multi-dimensional data collection and integration, the three types of data are organized in a unified time sequence to form the original potential fluctuation data, and a continuous potential change trajectory map is constructed through data domain classification and spatial coding.

3. The method of claim 2, wherein the method further comprises: In the process of creating the original draft of the potential floating record, the data collected from each energy storage cluster is archived in a unified timeline order, and the data records are simultaneously marked with the acquisition path identifier, time series label and spatial location code, so that the potential change trajectory of each energy storage cluster in the potential floating record has a complete physical correspondence.

4. The method of claim 2, wherein the method further comprises: The steps for cross bracing the net change curve are as follows: Based on the constructed potential floating original draft, the insulation resistance change trajectory of the energy storage cluster is extracted, and it is correlated item by item with the ground potential difference data in the potential floating original draft to establish the correspondence between potential change and insulation change. Based on the combined data sequence, the potential environment of the energy storage cluster is segmented and extracted to identify common potential drift factors and establish a potential influence spectrum. A correspondence was established between the group of factors to be stripped and the combined data sequence, and insulation baseline stripping was performed to remove common potential drift factors from the insulation resistance change trajectory, generating a net insulation change curve. The net insulation change curves are sequenced and labeled with status to form a data output structure with potential disturbance background markings.

5. The method of claim 4, wherein the method further comprises: When performing insulation baseline stripping, the potential drift characteristic segment constructed by the ground potential difference is used as the stripping reference. The potential offset influence of the corresponding time period is removed from the insulation resistance change trajectory item by item, so that the net insulation change curve after stripping only reflects the insulation degradation state of the energy storage cluster itself, and the sampling order and data field attribution are kept consistent in the data after stripping.

6. The method of claim 4, wherein the method further comprises: The steps for generating threshold isolation boundary parameters are as follows: A dynamic response range analysis framework is established around the net insulation change curve corresponding to each energy storage cluster. The net insulation resistance change data is divided into time periods and classified into states to construct a response range that matches the insulation degradation behavior of the energy storage cluster. Based on the obtained dynamic response range, configure a dynamic threshold range suitable for the characteristics of the energy storage cluster itself, and set upper and lower limit ranges with buffering characteristics. While constructing independent dynamic threshold ranges, the potential interference effects of adjacent energy storage clusters are recorded, and the interference information is appended to the threshold range configuration field of the energy storage cluster. The parameter fields corresponding to the energy storage clusters are summarized to form a cluster threshold mapping list, and the threshold isolation boundary parameters are formed by field fusion.

7. The method of claim 6, wherein the method further comprises: During the formation of threshold isolation boundary parameters, the cluster threshold mapping list is hierarchically labeled according to the potential interference level between energy storage clusters. Energy storage clusters with weak potential interference are labeled as allowable threshold adjustment areas, while energy storage clusters with strong potential interference are labeled as threshold restriction areas.

8. The method of claim 6, wherein the method further comprises: Threshold inheritance blocking control is implemented based on threshold isolation boundary parameters to restrict the cross-cluster threshold transfer paths in the cluster threshold mapping list, allowing only the threshold adjustment after potential disturbance assessment to enter the dynamic threshold range of the corresponding energy storage cluster. The steps are as follows: The threshold transfer paths between energy storage clusters are identified based on the generated cluster threshold mapping list, and the transfer paths are mapped to threshold isolation boundary parameters to construct a transfer path identification map; Based on the interference impact labels of the transmission path, the threshold adjustment permissions of the path are classified and screened, and threshold blocking control criteria are formulated to limit the threshold transmission range. For the state update requirements of energy storage clusters, a threshold adjustment screening operation based on the interference assessment results is performed, and only the threshold adjustment after potential interference assessment is allowed to enter the dynamic threshold range of the target energy storage cluster. The cluster threshold mapping list is updated around the adjusted dynamic threshold range, and the source of the transmission path, the level of interference impact, and the control status are recorded in the list.

9. The method of claim 8, wherein the method further comprises: When performing threshold adjustment filtering operations based on interference assessment results, the transmission path of each threshold adjustment is associated with the interference impact label, and an isolation mark is added to the blocked transmission path in the clustered threshold mapping list to continuously limit the threshold transmission of high interference paths in subsequent running cycles, ensuring the independent update and stability of the dynamic threshold range.

10. The method of claim 8, wherein the method further comprises: Based on the cluster threshold mapping list corrected by threshold inheritance blocking control, the following steps are taken to determine the insulation fault classification and early warning of each energy storage cluster under the constraint of potential floating recording band: Based on the cluster threshold mapping list corrected by threshold inheritance blocking control, an independent early warning judgment reference set is established for each energy storage cluster, and the dynamic threshold interval is associated with the potential fluctuation recording band. By constraining the potential environment of the energy storage cluster using the potential fluctuation recording band, the net insulation change curve is compared within the potential fluctuation range to determine the characteristics of the state change. Under the constraint of potential floating recording band, the insulation status of energy storage clusters is classified and determined according to the early warning reference set, and the position change of the net insulation change curve in the dynamic threshold interval is mapped to the early warning level. The time series processing and state preservation processing of the graded early warning results of energy storage clusters are carried out to prevent the occurrence of threshold reversal.

Images