Energy storage cabin terminal remote management method and system combined with distributed storage

Abstract

The application discloses a remote management method and system of energy storage cabin terminal combined with distributed storage, relates to the technical field of energy storage management, and comprises the following steps: sequentially deploying embedded storage nodes in M energy storage cabins to obtain M distributed storage nodes, and performing regional division to generate N cluster storage blocks; collecting and obtaining M node terminal data streams through the M distributed storage nodes, performing associated feature extraction on the M node terminal data streams to obtain M node working feature sets; performing fault identification on the M node working feature sets to determine M energy storage cabin fault feature sets; constructing an energy storage cabin strategy instruction library, performing control analysis on the M energy storage cabin fault feature sets, generating M energy storage cabin control instructions, and delivering the M energy storage cabin control instructions to the M energy storage cabins for terminal remote management. The application solves the technical problems of low remote management efficiency and untimely fault response of the prior art, and achieves the technical effects of improving the real-time performance of remote management and the fault processing efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage management, in particular to a remote management method and system for energy storage cabin terminals combined with distributed storage. BACKGROUND

[0002] As the core unit of large-scale energy storage systems, energy storage cabins are usually deployed in large numbers and widely distributed. When performing unified remote management on these energy storage cabins, it is necessary to collect and process the operation data of each cabin in real time, and quickly issue control instructions when an exception occurs. However, in actual applications, energy storage cabin data is usually scattered in each node, lacking a unified storage and processing mechanism, making it difficult to analyze the operation status information in a timely manner, and causing delays in control instruction responses, affecting the timeliness of fault handling and the overall operation efficiency of the system. SUMMARY

[0003] The present application provides a remote management method and system for energy storage cabin terminals combined with distributed storage, which is used to solve the technical problems of low efficiency and untimely fault response in the prior art.

[0004] In view of the above problems, the present application provides a remote management method and system for energy storage cabin terminals combined with distributed storage.

[0005] In a first aspect of the present application, a remote management method for energy storage cabin terminals combined with distributed storage is provided, which comprises:

[0006] M energy storage cabins are sequentially deployed with embedded storage nodes to obtain M distributed storage nodes, the M distributed storage nodes are regionally divided to generate N cluster storage blocks, wherein M≥N; M node terminal data streams are collected and obtained through the M distributed storage nodes, M node working feature sets are obtained by performing associated feature extraction on the M node terminal data streams; M energy storage cabin fault feature sets are determined based on the N cluster storage blocks for the M node working feature sets; an energy storage cabin strategy instruction library is constructed through a central cloud storage unit, control analysis is performed on the M energy storage cabin fault feature sets based on the energy storage cabin strategy instruction library to generate M energy storage cabin control instructions, and the M energy storage cabin control instructions are issued to the M energy storage cabins for terminal remote management.

[0007] In a second aspect of the present application, a remote management system for energy storage cabin terminals combined with distributed storage is provided, which comprises:

[0008] The regional division module is used for sequentially deploying embedded storage nodes in M energy storage compartments to obtain M distributed storage nodes, performing regional division on the M distributed storage nodes, and generating N cluster storage blocks, wherein M>N; the feature extraction module is used for collecting M node terminal data streams through the M distributed storage nodes, performing associated feature extraction on the M node terminal data streams, and obtaining M node working feature sets; the fault identification module is used for identifying faults of the M node working feature sets based on the N cluster storage blocks, determining M energy storage compartment fault feature sets; the remote management module is used for constructing an energy storage compartment strategy instruction library through a central cloud storage unit, performing control analysis on the M energy storage compartment fault feature sets based on the energy storage compartment strategy instruction library, generating M energy storage compartment control instructions, and issuing the M energy storage compartment control instructions to the M energy storage compartments for terminal remote management.

[0009] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0010] In the present application, embedded storage nodes are sequentially deployed in M energy storage compartments to obtain M distributed storage nodes, the M distributed storage nodes are regionally divided to generate N cluster storage blocks, wherein M>N; M node terminal data streams are collected through the M distributed storage nodes, associated feature extraction is performed on the M node terminal data streams to obtain M node working feature sets; faults of the M node working feature sets are identified based on the N cluster storage blocks to determine M energy storage compartment fault feature sets; an energy storage compartment strategy instruction library is constructed through a central cloud storage unit, control analysis is performed on the M energy storage compartment fault feature sets based on the energy storage compartment strategy instruction library to generate M energy storage compartment control instructions, and the M energy storage compartment control instructions are issued to the M energy storage compartments for terminal remote management. The present application solves the technical problems of low remote management efficiency and untimely fault response of the prior art, and achieves the technical effects of improving real-time remote management and fault processing efficiency by constructing distributed storage nodes and issuing centralized control instructions. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0012] Figure 1 The flowchart of the terminal remote management method of the energy storage compartment combined with distributed storage provided by the embodiments of the present application is shown.

[0013] Figure 2 A structure schematic diagram of a remote management system of an energy storage cabin terminal combined with distributed storage is provided in the embodiments of the present application.

[0014] The reference signs are explained as follows: a region division module 11, a feature extraction module 12, a fault identification module 13, and a remote management module 14. DETAILED DESCRIPTION

[0015] The present application provides a remote management method and system of an energy storage cabin terminal combined with distributed storage, aiming at solving the technical problems of low efficiency and untimely fault response of remote management of an energy storage cabin in the prior art, by constructing distributed storage nodes and issuing centralized fault identification and control instructions, the technical effects of improving real-time performance of remote management and fault processing efficiency are achieved.

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0017] It should be noted that any variation of the terms “comprise” and “have” is intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to the process, method, product or device.

[0018] Embodiment one, as shown in the present application provides a remote management method of an energy storage cabin terminal combined with distributed storage, which comprises: Figure 1

[0019] Step S100: sequentially deploying embedded storage nodes in M energy storage cabins to obtain M distributed storage nodes, performing region division on the M distributed storage nodes, and generating N cluster storage blocks, wherein M≥N.

[0020] In the embodiments of the present application, first, embedded storage nodes are sequentially deployed in M energy storage cabins, each embedded storage node collects and preliminarily processes the running data (such as voltage, current, temperature, health status, etc.) of a local energy storage system in real time by accessing the internal communication interface (such as CAN bus, Modbus or Ethernet) of the energy storage cabin. Through the deployment operation, M distributed storage nodes with local collection and processing capabilities are finally formed in the M energy storage cabins.

[0021] ​Next, the energy storage attributes of the M distributed storage nodes are analyzed, and key parameters including the energy storage system scale, performance requirements, and geographical location are extracted. Then, the target block quantity N is determined in combination with the energy storage cabin quantity and remote management requirements. Subsequently, the attribute parameters of each node are sorted according to priority to form an energy storage sequence parameter. Finally, the regional division is completed based on the sequence and the block quantity N to obtain N cluster storage blocks. Wherein, M≥N.

[0022] Further, the method provided by the application embodiment further comprises:

[0023] The energy storage attributes of the M distributed storage nodes are analyzed to obtain M node energy storage attribute parameters, including the energy storage system scale, performance requirements, and geographical location. The storage block quantity N is determined according to the energy storage cabin quantity and remote management requirements. The reference priority of each attribute parameter in the M node energy storage attribute parameters is sorted to obtain M node energy storage sequence parameters. The M node energy storage sequence parameters are regionally divided based on the storage block quantity N to obtain N cluster storage blocks.

[0024] In the application embodiment, first, the energy storage attributes of the M distributed storage nodes are analyzed to construct a complete node feature data set. This process collects and extracts three types of core parameters in real time through embedded storage nodes to form M node energy storage attribute parameters, including the energy storage system scale, performance requirements, and geographical location. The energy storage system scale is obtained by reading the configuration data in the battery management system (BMS), including the rated capacity (such as kWh) and the maximum charge and discharge power (such as kW) of each node corresponding to the energy storage cabin. The performance requirements such as maximum discharge power, response delay time, and cycle life indicators can be extracted by calling the device operation log or scheduling label. The geographical location is obtained by the embedded GPS module or the deployment map input information, which is usually represented in the form of latitude and longitude coordinates. In order to facilitate subsequent unified processing, the energy storage system scale and performance requirements are mapped to the [0, 1] interval through the minimum-maximum normalization method, and the geographical location is normalized based on the Euclidean distance between each node and the central control node to ensure the comparability of different dimension parameters. Finally, M node energy storage attribute parameters are formed.

[0025] Subsequently, the storage block quantity N is determined according to the energy storage cabin quantity and remote management requirements. The energy storage cabin quantity represents the number of entities participating in distributed management in the system, and the remote management requirements reflect the requirements of the control system on the division granularity, scheduling complexity, and parallel processing capability. For example, in the scenario of M=24 energy storage cabins, if the remote management system requires each management unit to control 4 to 6 cabins to ensure the scheduling accuracy and instruction feedback rate, the storage block quantity N can be reasonably set to 6.

[0026] Then, reference priority of each attribute in the M node energy storage attribute parameters is sorted to obtain M node energy storage sequence parameters. In the sorting process, a weighted scoring method is adopted. First, the normalized value of each attribute is taken as input, and a pre-set weight is adopted, for example, performance requirement weight is 0.5, energy storage system size is 0.3, and geographical position is 0.2. Then, the three parameter values of each node are weighted and summed to obtain the comprehensive score of the node. After the same processing is performed on the M nodes, the nodes are sorted from high to low according to the comprehensive score, and M node energy storage sequence parameters are obtained.

[0027] Finally, the M node energy storage sequence parameters are divided into regions based on the storage block quantity N. In this process, the M node energy storage sequence parameters are clustered based on the determined storage block quantity N. First, N cluster centers are initialized by random selection, and the Euclidean distance is taken as the measurement index to calculate the distance between each node and each cluster center. The M nodes are distributed to the nearest center to form N initial energy storage node clusters. Then, the cluster centers are adjusted based on the data mean value in each node cluster, and iteration optimization is performed until the cluster division is stable and no longer changes, and finally N energy storage optimized node clusters are obtained. The clustering result is taken as the N cluster storage blocks, and the structured grouping management of the M distributed storage nodes is realized.

[0028] Further, the method provided in the application embodiment further comprises the following steps.

[0029] The M node energy storage sequence parameters are randomly selected according to the storage block quantity N to initialize N initial cluster centers. The Euclidean distance calculation results of the M node energy storage sequence parameters and the N initial cluster centers are calculated respectively. The M node energy storage sequence parameters are distributed to the N initial cluster centers based on the Euclidean distance calculation results to obtain N initial energy storage node clusters. The cluster iteration optimization is performed based on the mean value information of the N initial energy storage node clusters to obtain N energy storage optimized node clusters, and the N energy storage optimized node clusters are taken as the N cluster storage blocks.

[0030] In the application embodiment, first, the M node energy storage sequence parameters are randomly selected according to the storage block quantity N, that is, N nodes are selected from the M nodes whose sorting has been completed by using a random sampling method without replacement, as the starting reference point of the current clustering process, that is, N initial cluster centers.

[0031] Then, the Euclidean distance calculation results of the M node energy storage sequence parameters and the N initial cluster centers are calculated respectively. In this process, the Euclidean distance is taken as the similarity measurement method in the feature space, which is suitable for the condition that the three-dimensional energy storage attributes have been normalized. For any node, the corresponding energy storage system size, performance requirement and geographical position are respectively denoted as , corresponding parameter of a certain initial cluster center , , the Euclidean distance calculation formula between the two is , D is the Euclidean distance calculation result. After performing the calculation on all M nodes and N initial centers, an MxN Euclidean distance matrix is obtained.

[0032] After obtaining the distance matrix, the energy storage sequence parameters of the M nodes are assigned to the N initial cluster centers based on the Euclidean distance calculation result, obtaining N initial energy storage node clusters. The assignment process follows the minimum distance attribution principle, that is, each node is assigned to the node cluster corresponding to the cluster center with the smallest Euclidean distance. The M nodes are sequentially traversed, and according to the minimum value index in the distance matrix, they are marked and classified into the corresponding initial cluster set. After this operation, all M nodes are divided into N non-overlapping initial node clusters, each cluster forms a preliminary aggregation around a center node in the attribute space, reflecting the initial attribute similarity structure between nodes. After this step, N preliminary formed energy storage node clusters are obtained.

[0033] Then, based on the mean value information of the N initial energy storage node clusters, the clustering iteration optimization is performed. This step constructs a new cluster center by calculating the average value of the energy storage system size, performance requirements and geographic location of all member nodes in each initial node cluster. Then, the Euclidean distance calculation is performed again for all nodes, and the node attribution is re-assigned based on the new distance relationship. This process continues to be iteratively executed until the node cluster division has no significant changes between consecutive two rounds, or the coordinate change amount of all cluster centers is less than the set convergence threshold. After this step, the converged N energy storage optimization node clusters are obtained, each cluster is internally close and externally discrete in the attribute space, and the division result is stable and reliable. Finally, the N energy storage optimization node clusters are stored as N cluster storage blocks.

[0034] Step S200: Collect M node terminal data streams through the M distributed storage nodes, extract the associated features of the M node terminal data streams, and obtain M node working feature sets.

[0035] In the embodiment of the present application, first, M node terminal data streams are collected by M distributed storage nodes. Each distributed storage node has an embedded data collection module integrated inside, adopts a multi-channel analog quantity collection method, and performs real-time sampling on the analog signals accessed through an ADC (analog-to-digital converter). The specific collection objects include battery cluster voltage, current, temperature, SOC (state of charge), and other operating parameters. Taking voltage collection as an example, a 10-time-per-second sampling operation is performed on the voltage input of each cell group, and a timestamp is recorded synchronously to generate a time sequence data frame with a unified format. These data are cached to a local ring buffer and then uploaded to a data management interface through Ethernet or a CAN bus to form continuous node operating data streams. After the process is completed, M node terminal data streams with consistent structure and unified format are obtained.

[0036] Subsequently, the M node terminal data streams are subjected to associated feature extraction, and the method adopted is a sliding time window feature extraction method. Specifically, the data stream of each node is taken as a unit, a fixed-length time window (such as 60 seconds) is set, and feature operations are performed on the collected multi-dimensional operating parameters in each time window. Taking SOC as an example, the mean value, change amplitude, and slope are calculated in each time window as a measure of energy change trend; for the temperature signal, the maximum temperature rise rate is calculated to identify the risk of overheating; and for the voltage signal, the maximum deviation value can be calculated to identify the consistency problem of the cell. All the calculation processes are based on the sliding operation of one-dimensional time sequence signals, and the statistical results in multiple time windows are aggregated into a feature vector by step extraction according to the window length.

[0037] Finally, the above sliding window feature extraction operation is independently performed on each distributed storage node to obtain M node operating feature sets.

[0038] Step S300: Based on the N cluster storage blocks, the M node operating feature sets are subjected to fault identification to determine M energy storage cabin fault feature sets.

[0039] In the embodiment of the present application, when the M node operating feature sets are subjected to fault identification based on the N cluster storage blocks, an energy storage cabin operating anomaly identifier is first constructed and stored in the N cluster storage blocks. Subsequently, the energy storage cabin operating anomaly identifier is called by each block to identify anomalies in the M node operating feature sets and extract M node anomaly feature sets. Finally, based on these anomaly feature sets, correlation analysis and fault prediction are carried out to identify potential risk patterns, and finally a structured M energy storage cabin fault feature set is generated.

[0040] Further, in the method provided by the embodiment of the present application, the determination of the M energy storage cabin fault feature sets further includes:

[0041] According to the energy storage cabin working standard, an energy storage cabin working abnormality identifier is constructed and stored in the N cluster storage blocks; the energy storage cabin working abnormality identifier is called through the N cluster storage blocks, the M node working feature sets are abnormally identified based on the energy storage cabin working abnormality identifier, and M node abnormal feature sets are obtained; the M node abnormal feature sets are associated and analyzed and fault prediction is performed, and M energy storage cabin fault feature sets are generated.

[0042] In the embodiments of the present application, first, an energy storage cabin working abnormality identifier is constructed according to the energy storage cabin working standard. The energy storage cabin working standard is pre-set by technical experts based on a large amount of energy storage system operation experience, and is used to define the reasonable range and fault boundary conditions of each operation index of the energy storage cabin. These standards clearly define the reference value and trigger logic of abnormality determination, for example, “single cell voltage deviation exceeds ± 50mV and duration exceeds 5 minutes” is set as consistency abnormality, and “temperature rise rate is higher than 5℃ / min and SOC change is less than 2%” is set as cooling efficiency abnormality. According to this, a set of abnormality judgment logic based on conditional expressions is constructed, and is packaged as a callable energy storage cabin working abnormality identifier. Each identification condition includes a judgment field, a threshold range, a judgment logic and an output result, and is accompanied by a pre-set abnormality level, such as slight, moderate or severe, for indicating different degrees of operation deviation risk. Then the energy storage cabin working abnormality identifier is deployed as a complete judgment set in the N cluster storage blocks, forming a local abnormality identification capability facing the partition.

[0043] Then the energy storage cabin working abnormality identifier is called through the N cluster storage blocks, and the M node working feature sets are abnormally identified based on the energy storage cabin working abnormality identifier. This step uses a rule matching method, taking the working features of each node as input, and performing logical comparison with the abnormal rules defined in the energy storage cabin working abnormality identifier one by one. The matching process is based on a Boolean condition judgment mechanism, for example, when a node has “voltage deviation of ± 62mV, temperature change rate of 6.2℃ / min, and SOC fluctuation of 1.1%” in a certain period, it is automatically identified that the node has triggered “cell consistency abnormality” and “cooling efficiency abnormality” at the same time. Each identification result is accompanied by an abnormal type, an abnormal level (such as “serious”), a trigger parameter and an occurrence time, and is uniformly recorded and output to the abnormal information set. This process is executed in parallel in the N cluster blocks, and finally M node abnormal feature sets are output.

[0044] Finally, the M node abnormal feature sets are analyzed and the faults are predicted. First, the abnormal features of each node are matched with the known fault types in the energy storage cabin fault mode library to obtain M node fault mode sets. Then, the influence of the current abnormality of each node on the system operation is evaluated to form M node fault level sets. Next, the fault mode sets are associated with rule mining and fault path analysis to identify possible fault evolution sequences and determine M energy storage cabin fault occurrence paths. Finally, the fault mode, level and path information are integrated to output the complete M energy storage cabin fault feature sets.

[0045] Further, the method provided by the application embodiment further comprises:

[0046] The energy storage cabin fault mode library is constructed, the M node abnormal feature sets are matched and analyzed based on the energy storage cabin fault mode library to obtain M node fault mode sets, the M node abnormal feature sets and the M node fault mode sets are evaluated in terms of influence degree to obtain M node fault level sets, the M node fault mode sets are associated with rule mining and fault path analysis to determine M energy storage cabin fault occurrence paths, and the M node fault mode sets, M node fault level sets and M energy storage cabin fault occurrence paths are used to generate the M energy storage cabin fault feature sets.

[0047] In the application embodiment, an energy storage cabin fault mode library is first constructed. This process uses a label classification method. Based on the energy storage system operation data collected in multiple actual operation projects, technical experts manually filter out typical abnormal states recorded in overcharging, over-discharging, temperature abnormality, communication loss and other scenarios, and further sort out the feature combinations when these abnormal states occur, for example, "the battery cell temperature difference is continuously higher than 8℃ for more than 10 minutes" is classified as "heat dissipation failure" mode, and "the voltage of a single string continuously drops by more than 10% and does not recover" is classified as "battery cell imbalance" mode. Each type of fault mode includes unique identification, trigger threshold group, key parameters, evolution trend description and other field information. After these mode data are collected and structured, a queryable and matchable energy storage cabin fault mode library is formed, which is used for subsequent fault feature identification.

[0048] Then, the M node abnormal feature sets are matched and analyzed based on the energy storage cabin failure mode library, and the field matching method is used. For the abnormal features of each distributed storage node, the trigger conditions in the failure mode library are compared with the features of the node one by one. For example, if the current temperature of node A is 57°C and the pressure difference is 150 mV, it is automatically searched whether there is a condition of "temperature > 55°C and pressure difference > 120 mV" in the mode. If it is satisfied, the node matches the failure mode. One node can correspond to multiple failure modes, and each matching relationship is assigned a matching degree value (for example, 3 out of 4 fields are satisfied, and the matching degree is 0.75). Finally, a set of matching results is extracted for each node to form a set of M node failure modes, which serves as the basis for preliminary diagnosis.

[0049] Subsequently, the influence degree of the M node abnormal feature set and the M node failure mode set is evaluated, and the threshold weighted scoring method is adopted. Each type of abnormal feature is preset with a weight factor (for example, the weight of temperature rise rate is 0.4, and the weight of current fluctuation is 0.3), and a scoring interval is defined (for example, a rate higher than 3°C / min is 60 points, and a rate higher than 5°C / min is 80 points). According to the actual abnormal feature value of each node, each index is scored, and the total influence score is calculated according to the preset weight. For example, the weighted abnormal score of a node is 72 points, and according to the set grade division standard (0-40 is slight, 41-70 is moderate, and 71 and above is severe), the node is rated as "severe" impact level. After all nodes complete the evaluation, a set of M node failure level sets is formed.

[0050] Subsequently, the M node failure mode set is subjected to association rule mining and failure path analysis. In this process, a confidence threshold is first set to screen potential failure mode combinations among nodes, and a failure rule association network reflecting the causal relationship between modes is constructed. Subsequently, based on the frequency, order and influence weight of each mode node in each path, the criticality is evaluated and a set of failure path criticality is formed. Finally, the most representative and early warning value path of each node is identified based on the criticality information, and the failure occurrence path of the M energy storage cabin is determined.

[0051] Finally, based on the M node failure mode set, the M node failure level set and the M energy storage cabin failure occurrence path, information integration processing is performed. This process uses the field merging method to uniformly collect the node failure mode information, failure level information and key node information involved in the failure occurrence path corresponding to each energy storage cabin. Specifically, it includes node identification, corresponding failure mode type, failure level result and the impact position of the failure in the path. All the summary results are stored in the failure feature data set according to the unified field structure, thereby constructing a set of M energy storage cabin failure features.

[0052] Further, the method provided by the application embodiment further comprises the following steps of:

[0053] a preset confidence threshold is set, the M node fault mode sets are associated rule mined according to the confidence threshold, M fault rule association networks are generated, each fault path in the M fault rule association networks is key importance evaluated and summed, and M fault path key degree sets are obtained, the M fault path key degree sets are used for key path identification of the M fault rule association networks, and the M energy storage cabin fault occurrence paths are determined.

[0054] In the application embodiment, firstly, a confidence threshold is set as a minimum credible basis for the existence of an associated relationship between fault modes. For example, the confidence threshold is set to 0.6, which means that only fault mode combinations with at least 60% conditional probability in the co-occurrence of abnormal states are retained. Subsequently, a frequent item set mining method is used to analyze each of the M node fault mode sets one by one, the frequency of co-occurrence of various fault mode combinations in the M nodes is counted, and the support degree and the confidence degree of each mode combination are calculated. Taking the combination of “cell imbalance” and “temperature anomaly” as an example, if the combination co-occurs 30 times in the M nodes, and “cell imbalance” independently occurs 50 times, the confidence degree of the combination is 30 / 50 = 0.6, which meets the set condition. All mode combinations that meet the confidence threshold are organized as directed edge relationships, indicating causal linkage, and M corresponding fault rule association networks are constructed.

[0055] After the fault rule association network is completed, the fault path weighted accumulation method is used to evaluate the key importance of each path in the M networks. Firstly, the fault levels (minor, moderate, and severe) are mapped to fixed weight values with reference to the aforementioned M node fault level sets, for example, minor is 1, moderate is 2, and severe is 3. For all paths in each node network, the level weights of the fault mode nodes in the path are extracted and summed. For example, if the fault levels involved in a path “communication anomaly → voltage fluctuation → cell imbalance” are moderate, minor, and severe, respectively, the path key degree is 2+1+3 = 6. This calculation process is performed on the M nodes one by one, and finally M fault path key degree sets are formed, each set reflecting the influence strength of all possible fault paths in a node.

[0056] After obtaining the M sets of fault path criticality, critical path identification is performed, and the maximum criticality priority selection method is used for screening. Specifically, in the fault rule association network corresponding to each node, the path with the highest criticality score is selected as the critical fault path of the node. When there are multiple paths with the same criticality, the path with the shortest length is preferentially selected; if the lengths are still the same, the first fault mode in the path is further compared, and the path with a higher level is preferentially selected. This method takes into account the influence strength and urgency of the fault chain, and finally determines M fault occurrence paths of the M energy storage tanks.

[0057] Step S400: Constructing an energy storage tank strategy instruction library through a central cloud storage unit, controlling and analyzing the M sets of energy storage tank fault feature sets based on the energy storage tank strategy instruction library, generating M energy storage tank control instructions, and issuing the M energy storage tank control instructions to the M energy storage tanks for terminal remote management.

[0058] In the embodiments of the present application, an energy storage tank strategy instruction library is first constructed through a central cloud storage unit. This process uses a strategy classification method, and technical experts classify and structure the response measures taken in historical energy storage tank fault events based on the control response schemes summarized in multiple actual operation and maintenance projects. Specifically, it includes extracting fault types, response logic, execution conditions, instruction content, and priority, and setting standard strategy templates for common faults such as "voltage imbalance", "battery cell temperature rise", "communication interruption", etc. The structured strategy information is stored in the central cloud storage unit in a structured tag form (such as JSON or SQL format), supporting quick retrieval and dynamic update according to fault tags. After this step, an energy storage tank strategy instruction library covering common fault scenarios is generated, which can be matched and called.

[0059] Next, the M sets of energy storage tank fault feature sets are controlled and analyzed based on the energy storage tank strategy instruction library, and the fault strategy instructions corresponding to each energy storage tank are obtained through index matching method. Further integrated analysis is performed based on the fault correlation between the energy storage tanks, the global fault root cause is identified, and the initial strategy instructions are modified accordingly, so as to generate M energy storage tank control instructions with better coordination and execution efficiency.

[0060] Finally, the M energy storage tank control instructions are issued to the M energy storage tanks for terminal remote management. In this process, the M energy storage tank control instructions are encrypted and issued in the form of protocol encapsulation. After each energy storage tank terminal receives the encrypted control instruction, it completes the local decryption operation and executes the corresponding remote management action, and at the same time, the execution status is fed back to the central platform, completing the closed-loop control process and realizing the terminal remote management of the M energy storage tanks.

[0061] Further, the method provided by the application embodiment further comprises the following steps.

[0062] The M energy storage cabin fault feature sets are subjected to index matching based on the energy storage cabin strategy instruction library to obtain M fault strategy instructions; the M energy storage cabin fault feature sets are subjected to correlation integration analysis to determine an energy storage cabin global fault root cause; and the M fault strategy instructions are subjected to collaborative correction based on the energy storage cabin global fault root cause to generate the M energy storage cabin control instructions.

[0063] In the application embodiment, first, the fault features of each energy storage cabin are quickly matched based on the energy storage cabin strategy instruction library constructed in the central cloud storage unit. Specifically, by means of an index matching method, the field tags (such as fault type, voltage offset, temperature rise rate, and communication state) in each energy storage cabin fault feature set are one-to-one compared with the preset response rules in the energy storage cabin strategy instruction library to filter out the coping strategies meeting the conditions. For example, if the "voltage difference higher than 120 mV" and "single cell temperature rise rate higher than 5 ℃ / min" are recorded in a certain energy storage cabin feature set, the preset strategy template of "triggering cell balancing + accelerating the air cooling system" is matched. After this process, M fault strategy instructions corresponding to the M energy storage cabins are extracted respectively.

[0064] Next, the M energy storage cabin fault feature sets are subjected to correlation integration analysis to identify the common fault causes in the system range. This step adopts a sliding window clustering analysis method. First, a unified time sliding window parameter (for example, each sliding window length is 5 minutes and the sliding interval is 1 minute) is set, and the fault events reported by the M energy storage cabin nodes in each sliding window are grouped and aggregated in time sequence. Each fault event is represented by a structured feature vector, and the feature fields include "node ID", "voltage offset value", "temperature rise rate", "signal interruption state", etc., to form a feature vector group. Then, the feature vectors of any two energy storage cabin nodes in the sliding window are subjected to pairwise similarity measurement by means of a cosine similarity calculation method to determine whether they present similar abnormal trends in the same time period. If the similarity exceeds a set threshold (for example, 0.9), it is considered as the same type of event. The highly similar fault events are classified by means of a hierarchical clustering method, and the fault mode set with the highest occurrence frequency and the most involved nodes is identified as the energy storage cabin global fault root cause in the current time window. For example, when multiple energy storage cabins simultaneously present heat dissipation failure phenomenon in the same time period, it is determined that "the performance degradation of the whole cabinet air cooling system" is the energy storage cabin global fault root cause in the current cycle.

[0065] Finally, based on the identified global fault root cause of the energy storage cabin, the extracted M fault strategy instructions are cooperatively corrected. The method for optimizing and adjusting the aforementioned M fault strategy instructions is a strategy priority cooperative correction mechanism. Each fault strategy instruction contains response priority, control parameters, and execution dependent resources in the energy storage cabin strategy instruction library. When it is found that some strategies are highly related to the currently identified global root cause (such as multiple nodes triggering cooling response strategies due to abnormal temperature rise), the priority of these strategies is raised to the highest level; at the same time, the strategies unrelated or secondary to the global root cause are downgraded or postponed for execution. If there are multiple strategies calling the same resource (such as multiple energy storage cabins requesting PCS module power at the same time), the resource conflict is resolved through priority sorting and polling scheduling to ensure the coordination and consistency between control instructions. After the above index matching, global fault root cause identification, and priority correction, M updated energy storage cabin control instructions are generated for the M energy storage cabins. Each control instruction explicitly includes target node ID, execution instruction content (such as "start air cooling unit" and "start cell balancing"), execution parameters (such as "cooling intensity = 80%"), and scheduling priority.

[0066] Further, the method provided by the application embodiment further comprises:

[0067] The M energy storage cabin control instructions are encrypted, and the encrypted M energy storage cabin control instruction protocols are issued to the M energy storage cabins; the M energy storage cabins perform terminal receiving decryption and remote execution feedback on the encrypted M energy storage cabin control instructions.

[0068] In the application embodiment, in order to ensure the communication security and instruction accuracy of the M energy storage cabin control instructions during the issuing and execution process, the M energy storage cabin control instructions are first subjected to encryption processing respectively. Specifically, for each energy storage cabin control instruction, the instruction content is encrypted using the Advanced Encryption Standard (AES-256). The encryption input includes the target energy storage cabin identifier corresponding to the energy storage cabin control instruction, control parameters (such as air cooling intensity, cell balancing threshold, power scheduling amount, etc.), timestamp, and instruction priority field, and an initialization vector is generated by combining the unique key index stored locally by the energy storage cabin, and the encryption of the M energy storage cabin control instructions is completed through a key encryption logic module.

[0069] After encryption, the encrypted M energy storage cabin control instructions are encapsulated using a protocol structure (such as Modbus TCP or IEC 61850) conforming to the industrial communication standard to construct M communication messages. Each communication message contains an energy storage cabin unique identifier, a key index, an encrypted instruction field, a timestamp verification information, and a redundancy check bit, ensuring that the data packet has identity confirmation and transmission integrity. Subsequently, the M communication messages are transmitted in parallel to the corresponding M energy storage cabins through an industrial communication link (such as a 5G public network dedicated line, a LoRa ad hoc network, or an industrial Ethernet), realizing the remote delivery of encrypted instructions.

[0070] After receiving the respective encrypted instruction messages, the corresponding M energy storage cabins first retrieve the matching key from the local key management structure based on the energy storage cabin identifier and the key index, perform AES-256 decryption on the received encrypted instruction field, and restore the original energy storage cabin control instruction content. After successful decryption, the energy storage cabin starts the local execution process based on the control parameters and execution conditions in the instruction, and performs specific control operations on related devices (such as PCS, battery management unit, cooling unit, etc.).

[0071] After completing the execution operation, each energy storage cabin generates a remote feedback data packet containing the execution result, execution status code, and feedback timestamp, and returns it uplink through the same communication path as the receiving link. The central processing structure receives the remote feedback returned by the M energy storage cabins, records the control response status, and forms a closed-loop control log. Thus, the entire process of encryption, protocol transmission, terminal reception and decryption, and remote execution feedback of the M energy storage cabin control instructions is completed.

[0072] Further, the method provided by the application embodiment further comprises:

[0073] When the M energy storage cabin control instructions are upgrade instructions, the M energy storage cabins receive the upgrade package in blocks according to the YMODEM protocol and perform CRC verification and breakpoint resume processing on the upgrade package data.

[0074] In the application embodiment, when the M energy storage cabin control instructions are upgrade instructions, the M energy storage cabins will start the firmware upgrade process in turn. Each energy storage cabin receives the upgrade package sent by the control end through serial communication according to the YMODEM protocol. The YMODEM protocol supports dividing the entire upgrade package into multiple fixed-size data blocks (e.g., 1024 bytes), and the energy storage cabin receives the data blocks one by one and performs CRC verification on each data block to confirm the completeness of the data. If a data block fails the verification, it will automatically request retransmission of the data block.

[0075] In the transmission process, if the network is interrupted or the device is abnormal, causing the upgrade to be interrupted, the energy storage cabin will record the last complete data block number received at present, and after recovery, the connection is re-established, and the receiving is continued from the interrupted position to complete the breakpoint resume processing.

[0076] In the embodiments of the present application, as described above, the embodiments of the present application have at least the following technical effects:

[0077] The present application sequentially deploys embedded storage nodes in M energy storage cabins to obtain M distributed storage nodes, divides the M distributed storage nodes into N cluster storage blocks, wherein M > N; collects M node terminal data streams through the M distributed storage nodes, extracts associated features of the M node terminal data streams to obtain M node working feature sets; identifies faults of the M node working feature sets based on the N cluster storage blocks to determine M energy storage cabin fault feature sets; constructs an energy storage cabin strategy instruction library through a central cloud storage unit, controls and analyzes the M energy storage cabin fault feature sets based on the energy storage cabin strategy instruction library to generate M energy storage cabin control instructions, and issues the M energy storage cabin control instructions to the M energy storage cabins for terminal remote management. The present application solves the technical problems of low remote management efficiency and untimely fault response of the existing technology, and achieves the technical effects of improving the real-time performance of remote management and the efficiency of fault processing by constructing distributed storage nodes and issuing centralized fault identification and control instructions.

[0078] Embodiment two, based on the same inventive concept as the terminal remote management method of the energy storage cabin combined with distributed storage in the foregoing embodiments, as shown in Figure 2 The present application provides a terminal remote management system of an energy storage cabin combined with distributed storage, and the system and method embodiments in the embodiments of the present application are based on the same inventive concept. The system comprises:

[0079] The regional division module 11 is configured to sequentially deploy embedded storage nodes in M energy storage cabins to obtain M distributed storage nodes, divide the M distributed storage nodes into N cluster storage blocks, and M≥N; the feature extraction module 12 is configured to collect M node terminal data streams through the M distributed storage nodes, extract associated features of the M node terminal data streams, and obtain M node working feature sets; the fault identification module 13 is configured to identify faults of the M node working feature sets based on the N cluster storage blocks, and determine M energy storage cabin fault feature sets; and the remote management module 14 is configured to construct an energy storage cabin strategy instruction library through a central cloud storage unit, control and analyze the M energy storage cabin fault feature sets based on the energy storage cabin strategy instruction library, generate M energy storage cabin control instructions, and issue the M energy storage cabin control instructions to the M energy storage cabins for terminal remote management.

[0080] Further, the system is also configured to implement the following functions:

[0081] The energy storage attribute of the M distributed storage nodes is analyzed to obtain M node energy storage attribute parameters, the M node energy storage attribute parameters including energy storage system scale, performance requirements, and geographical location; the number N of storage blocks is determined according to the number of energy storage cabins and remote management requirements; the reference priority of each attribute parameter in the M node energy storage attribute parameters is sorted to obtain M node energy storage sequence parameters; and the M node energy storage sequence parameters are divided into N cluster storage blocks based on the number N of storage blocks.

[0082] Further, the system is also configured to implement the following functions:

[0083] The M node energy storage sequence parameters are randomly selected according to the number N of storage blocks to initialize N initial cluster centers; the Euclidean distance calculation results of the M node energy storage sequence parameters and the N initial cluster centers are calculated respectively; the M node energy storage sequence parameters are assigned to the N initial cluster centers based on the Euclidean distance calculation results to obtain N initial energy storage node clusters; the N initial energy storage node clusters are used to perform clustering iteration optimization based on the mean information of the N initial energy storage node clusters to obtain N energy storage optimized node clusters, and the N energy storage optimized node clusters are used as the N cluster storage blocks.

[0084] Further, the system is also configured to implement the following functions:

[0085] According to the energy storage cabin working standard, an energy storage cabin working abnormality identifier is constructed and stored in the N cluster storage blocks; the energy storage cabin working abnormality identifier is called through the N cluster storage blocks, the M node working feature sets are abnormally identified based on the energy storage cabin working abnormality identifier, and M node abnormal feature sets are obtained; the M node abnormal feature sets are analyzed and predicted for faults to generate M energy storage cabin fault feature sets.

[0086] Further, the system is also used to implement the following functions:

[0087] An energy storage cabin fault mode library is constructed, the M node abnormal feature sets are matched and analyzed based on the energy storage cabin fault mode library, M node fault mode sets are obtained; the M node abnormal feature sets and the M node fault mode sets are evaluated for influence degree to obtain M node fault level sets; the M node fault mode sets are analyzed for association rules and fault paths to determine M energy storage cabin fault occurrence paths; based on the M node fault mode sets, the M node fault level sets and the M energy storage cabin fault occurrence paths, the M energy storage cabin fault feature sets are generated.

[0088] Further, the system is also used to implement the following functions:

[0089] A confidence threshold is preset, the M node fault mode sets are analyzed for association rules according to the confidence threshold to generate M fault rule association networks; each fault path in the M fault rule association networks is evaluated for criticality and summed to obtain M fault path criticality sets; the M fault rule association networks are identified for critical paths based on the M fault path criticality sets to determine the M energy storage cabin fault occurrence paths.

[0090] Further, the system is also used to implement the following functions:

[0091] Based on the energy storage cabin strategy instruction library, the M energy storage cabin fault feature sets are index matched to obtain M fault strategy instructions; the M energy storage cabin fault feature sets are analyzed for association integration to determine an energy storage cabin global fault root cause; based on the energy storage cabin global fault root cause, the M fault strategy instructions are cooperatively corrected to generate the M energy storage cabin control instructions.

[0092] Further, the system is also used to implement the following functions:

[0093] The M energy storage cabin control instructions are encrypted, and the encrypted M energy storage cabin control instruction protocols are issued to the M energy storage cabins; the encrypted M energy storage cabin control instructions are received, decrypted and remotely executed and fed back by the M energy storage cabins.

[0094] Further, the system is also used to realize the following functions:

[0095] When the M energy storage cabin control instruction is an upgrade instruction, the M energy storage cabin receives the upgrade package in blocks according to the YMODEM protocol and performs upgrade package data CRC check and breakpoint resume processing.

[0096] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes a specific embodiment of the present application. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0097] The above only describes the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0098] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application intends to include these modifications and changes.

Claims

1. A method for remote management of an energy storage cabin terminal combined with distributed storage, characterized in that, The method comprises: sequentially deploying embedded storage nodes in M energy storage compartments to obtain M distributed storage nodes, performing regional division on the M distributed storage nodes to generate N cluster storage blocks, wherein M≥N; collecting M node terminal data streams through the M distributed storage nodes, performing associated feature extraction on the M node terminal data streams to obtain M node working feature sets; performing fault identification on the M node working feature sets based on the N cluster storage blocks to determine M energy storage compartment fault feature sets; constructing an energy storage compartment strategy instruction library through a central cloud storage unit, performing control analysis on the M energy storage compartment fault feature sets based on the energy storage compartment strategy instruction library to generate M energy storage compartment control instructions, and issuing the M energy storage compartment control instructions to the M energy storage compartments for terminal remote management; wherein the generation of N cluster storage blocks comprises: performing energy storage attribute analysis on the M distributed storage nodes to obtain M node energy storage attribute parameters, the M node energy storage attribute parameters including energy storage system size, performance requirements, and geographical location; determining the number N of storage blocks according to the number of energy storage compartments and remote management requirements; performing reference priority sorting on each attribute parameter in the M node energy storage attribute parameters to obtain M node energy storage sequence parameters; performing regional division on the M node energy storage sequence parameters based on the number N of storage blocks to obtain N cluster storage blocks; wherein the determination of M energy storage compartment fault feature sets comprises: constructing an energy storage compartment working abnormality identifier according to energy storage compartment working standards, and storing the energy storage compartment working abnormality identifier in the N cluster storage blocks; calling the energy storage compartment working abnormality identifier through the N cluster storage blocks, performing abnormality identification on the M node working feature sets based on the energy storage compartment working abnormality identifier to obtain M node abnormality feature sets; performing correlation analysis and fault prediction on the M node abnormality feature sets to generate M energy storage compartment fault feature sets; wherein the generation of M energy storage compartment control instructions comprises: performing index matching on the M energy storage compartment fault feature sets based on the energy storage compartment strategy instruction library to obtain M fault strategy instructions; performing associated integrated analysis on the M energy storage compartment fault feature sets to determine an energy storage compartment global fault root cause; performing collaborative correction on the M fault strategy instructions based on the energy storage compartment global fault root cause to generate the M energy storage compartment control instructions.

2. The method for remote management of an energy storage hutch terminal integrated with distributed storage according to claim 1, wherein, The obtaining of N cluster storage blocks comprises: randomly selecting the M node energy storage sequence parameters according to the number N of storage blocks to initialize N initial cluster centers; calculating the Euclidean distance calculation results of the M node energy storage sequence parameters and the N initial cluster centers, respectively; based on the Euclidean distance calculation results, assigning the M node energy storage sequence parameters to the N initial cluster centers to obtain N initial energy storage node clusters. Performing clustering iterative optimization based on the mean information of the N initial energy storage node clusters, obtaining N energy storage optimization node clusters, and storing the N energy storage optimization node clusters as the N cluster storage blocks.

3. The method for remote management of an energy storage hutch terminal integrated with distributed storage of claim 1, wherein, The generating of the M energy storage cabin fault feature sets comprises: Constructing an energy storage cabin fault mode library, performing matching analysis on the M node anomaly feature sets based on the energy storage cabin fault mode library, and obtaining M node fault mode sets; Performing influence degree evaluation on the M node anomaly feature sets and the M node fault mode sets, and obtaining M node fault level sets; Performing association rule mining and fault path analysis on the M node fault mode sets, and determining M energy storage cabin fault occurrence paths; Based on the M node fault mode sets, M node fault level sets and the M energy storage cabin fault occurrence paths, the M energy storage cabin fault feature sets are generated.

4. The method for remote management of an energy storage hutch terminal integrated with distributed storage of claim 3, wherein, The determination of the M energy storage cabin fault occurrence paths comprises: Pre-setting a confidence threshold, performing association rule mining on the M node fault mode sets according to the confidence threshold, and generating M fault rule association networks; Performing criticality evaluation and summation calculation on each fault path in the M fault rule association networks, and obtaining M fault path criticality sets; Based on the M fault path criticality sets, the M fault rule association networks are identified for critical paths, and the M energy storage cabin fault occurrence paths are determined.

5. The method for remote management of an energy storage hutch terminal integrated with distributed storage of claim 1, wherein, The issuing of the M energy storage cabin control instructions to the M energy storage cabins for terminal remote management comprises: Encrypting the M energy storage cabin control instructions, and issuing the encrypted M energy storage cabin control instruction protocols to the M energy storage cabins; Through the M energy storage cabins, the encrypted M energy storage cabin control instructions are received, decrypted and remotely executed for feedback.

6. The method for remote management of an energy storage hutch terminal integrated with distributed storage of claim 5, wherein, The method further comprises: When the M energy storage cabin control instructions are upgrade instructions, the M energy storage cabins receive the upgrade packages in blocks according to the YMODEM protocol, and perform upgrade package data CRC check and breakpoint resume processing.

7. The remote management system of the energy storage cabin terminal combined with distributed storage, characterized in that, The system is used to perform the terminal remote management method of the energy storage cabin combined with distributed storage as claimed in any one of claims 1-6, and the system comprises: A region division module is used to sequentially deploy embedded storage nodes in M energy storage cabins to obtain M distributed storage nodes, divide the M distributed storage nodes into regions, and generate N cluster storage blocks, wherein M≥N; A feature extraction module is used to collect M node terminal data streams through the M distributed storage nodes, extract associated features from the M node terminal data streams, and obtain M node working feature sets; A fault identification module is used to identify faults based on the N cluster storage blocks, and determine M energy storage cabin fault feature sets; A remote management module is used to construct an energy storage cabin strategy instruction library through a central cloud storage unit, perform control analysis on the M energy storage cabin fault feature sets based on the energy storage cabin strategy instruction library, generate M energy storage cabin control instructions, and issue the M energy storage cabin control instructions to the M energy storage cabins for terminal remote management.

Images