CN-121979888-A - Archiving method, equipment and storage medium for life cycle data of energy storage battery cell

CN121979888ACN 121979888 ACN121979888 ACN 121979888ACN-121979888-A

Abstract

An archiving method, equipment and storage medium for life cycle data of an energy storage battery cell relate to the technical field of battery management; according to the archiving method, the archiving equipment and the storage medium for the life cycle data of the energy storage battery core, firstly, in the real-time operation process of the battery core, according to the preset evaluation criteria, the original operation data time period with the preset information value is dynamically identified, the feature vector is extracted, then the event node with obvious transition is detected and structured and recorded based on the feature vector, and finally, the refined life cycle data archive of the battery core is integrated and built, so that the transition from low-value original data flow to a high-density feature knowledge base is realized, and meanwhile, the high-value data base with clear structure and direct utilization is provided for battery core health state analysis, evolution track tracing and intelligent prediction.

Inventors

FENG GUIQING
ZHANG XINYONG
HU ZHIHAN
ZHANG LI
WANG LIANGLIANG
LIU JIANFENG

Assignees

深能源(深圳)创新技术有限公司
深圳市盛路物联通讯技术有限公司
深能北方能源控股有限公司

Dates

Publication Date: 20260505
Application Date: 20260125

Claims (10)

1. An archiving method of life cycle data of an energy storage battery cell, comprising the following steps: In the real-time operation process of the battery cell, dynamically identifying an original operation data period with preset information value according to a preset evaluation criterion; performing dimension reduction feature extraction on the identified time period of the original operation data, generating a feature vector representing the behavior feature of the battery cell in the time period, and storing the feature vector in association with metadata reflecting the acquisition background of the time period; based on a time sequence evolution rule of the continuously stored feature vector, detecting event nodes representing significant transition of the health state of the battery cell, and generating a structured event record containing event features, state comparison before and after transition and associated feature vectors for each detected event node; and integrating and correlating the accumulated and stored feature vector set with the structured event record sequence to construct a refined life cycle data file of the battery cell.
2. The method of claim 1, wherein dynamically identifying the period of raw operational data having a predetermined information value based on a predetermined evaluation criterion comprises at least one of: Determining whether the data period has a preset information value based on whether the change characteristics of the key operation parameters of the battery cells in the data period exceed a dynamic reference range established based on historical operation data of the battery cells; Determining whether a corresponding operation data period has a preset information value or not based on whether the battery core enters a preset specific key working condition interval, wherein the specific key working condition interval is associated with a typical process of aging of a battery core material, interface side reaction or evolution of a safety state; determining whether the data period has a preset information value based on whether a time interval between an acquisition time point of the data period and an occurrence time of a preset key operation event or an external environment threshold event in a battery life cycle is within a preset time correlation window.
3. The method according to claim 2, characterized in that said dimension-reduction feature extraction of said identified periods of raw operational data is performed using a neural network based feature encoder; The feature encoder is obtained by performing self-supervision pre-training on massive battery cell historical operation data, and the encoded output is a low-dimensional feature vector capable of reconstructing primary mode information of original data.
4. The method of claim 3, wherein the metadata reflecting the context of the time period acquisition includes at least a value determination condition identification triggering data identified as valuable for the time period, a time period start-stop timestamp, an operating condition classification tag for the battery cell during the time period, and an index identification of a preceding and following data time period adjacent to the time period in a time sequence.
5. The method of claim 1, wherein detecting an event node that characterizes a significant transition in a state of health of a cell comprises: calculating at least one health status derived indicator based on the sequence of feature vectors; executing an online variable point detection algorithm on the time sequence of the health state derivative index, and identifying candidate time points at which the index statistical characteristics are mutated; Comparing the statistical distribution of the feature vector sets in the front and back preset time windows of each candidate time point; If the distribution difference passes the preset significance test, judging the candidate time point as a confirmation event node representing the significant transition of the health state.
6. The method of claim 5, wherein in the structured event record generated for the detected event node, the event characteristic is selected from a predefined set of event type labels, the label definition in the set of event type labels is determined based on a combination of information in at least two dimensions including a change pattern characteristic of a characteristic vector, an affected health-derived index type, and an operating condition change of the cell before and after the event.
7. The method according to claim 1, wherein the method further comprises: And uploading all or part of the content of the refined life cycle data file to a remote analysis platform according to a preset transmission strategy, wherein the triggering condition of the transmission strategy comprises at least one of periodic time triggering, triggering when the utilization rate of the local storage resources reaches a threshold value, and triggering in response to a data file inquiry request aiming at a specific event type or a specific time range and issued by the remote analysis platform.
8. An archive device for storing life cycle data of an electric core, comprising: The identification module is used for dynamically identifying the original operation data time period with the preset information value according to the preset evaluation criterion in the real-time operation process of the battery cell; The association storage module is used for carrying out dimension reduction feature extraction on the identified time period of the original operation data, generating a feature vector representing the behavior feature of the battery cell in the time period, and associating and storing the feature vector with metadata reflecting the acquisition background of the time period; The detection module is used for detecting event nodes representing significant transition of the health state of the battery cell based on a time sequence evolution rule of the continuously stored feature vectors, and generating a structured event record containing event features, state comparison before and after transition and associated feature vectors for each detected event node; and the construction module is used for integrating and correlating the accumulated and stored feature vector set with the structured event record sequence to construct a refined life cycle data file of the battery cell.
9. An archiving device for storing life cycle data of a battery cell, comprising a processor, a memory and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when the computer program is executed by the processor.
10. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the method according to any one of claims 1 to 7.

Description

Archiving method, equipment and storage medium for life cycle data of energy storage battery cell Technical Field The application belongs to the technical field of battery management, and particularly relates to a method, equipment and a storage medium for archiving life cycle data of an energy storage battery core. Background The existing energy storage battery core life cycle data management mainly adopts a full archiving mode, and massive original operation data are stored indiscriminately. The method has the advantages that the method has low storage value density, a large amount of redundant low-value steady-state data occupy storage resources, a high-value data period which really contains state transition information is submerged, and the method has insufficient analysis efficiency, and is characterized by disjointing the morphology between the archived original data stream and macroscopic state characteristics and key evolution events required by analysis, so that the subsequent state evaluation, life prediction and fault tracing analysis efficiency is low, and the calculation is complex. Therefore, there is a need for an efficient archiving method that can achieve data quality improvement and reduction from the source and directly output analysis-friendly knowledge. Disclosure of Invention In view of the above, the embodiments of the present application provide a method, an apparatus, and a storage medium for archiving life cycle data of an energy storage battery cell, which aim to greatly improve the storage efficiency of the battery cell data and directly provide a high-value data base with clear structure and direct utilization for battery cell health state analysis, evolution track tracing, and intelligent prediction. The embodiment of the application provides an archiving method of life cycle data of an energy storage battery cell, comprising the following steps: In the real-time operation process of the battery cell, dynamically identifying an original operation data period with preset information value according to a preset evaluation criterion; performing dimension reduction feature extraction on the identified time period of the original operation data, generating a feature vector representing the behavior feature of the battery cell in the time period, and storing the feature vector in association with metadata reflecting the acquisition background of the time period; based on a time sequence evolution rule of the continuously stored feature vector, detecting event nodes representing significant transition of the health state of the battery cell, and generating a structured event record containing event features, state comparison before and after transition and associated feature vectors for each detected event node; and integrating and correlating the accumulated and stored feature vector set with the structured event record sequence to construct a refined life cycle data file of the battery cell. In an embodiment, the dynamically identifying the period of the original operation data having the predetermined information value according to the predetermined evaluation criteria includes at least one of the following: Determining whether the data period has a preset information value based on whether the change characteristics of the key operation parameters of the battery cells in the data period exceed a dynamic reference range established based on historical operation data of the battery cells; Determining whether a corresponding operation data period has a preset information value or not based on whether the battery core enters a preset specific key working condition interval, wherein the specific key working condition interval is associated with a typical process of aging of a battery core material, interface side reaction or evolution of a safety state; determining whether the data period has a preset information value based on whether a time interval between an acquisition time point of the data period and an occurrence time of a preset key operation event or an external environment threshold event in a battery life cycle is within a preset time correlation window. In an embodiment, the dimension reduction feature extraction is performed on the identified original operation data period, and is specifically implemented by a feature encoder based on a neural network; The feature encoder is obtained by performing self-supervision pre-training on massive battery cell historical operation data, and the encoded output is a low-dimensional feature vector capable of reconstructing primary mode information of original data. In one embodiment, the metadata reflecting the period acquisition context at least comprises a value judgment condition identifier for triggering the data of which the period is identified as valuable, a period starting and ending time stamp, an operation condition classification label of the battery cell in the period and an index identifier of a front data period and a b