CN-121997273-A - Dynamic control method of lithium battery energy storage system

CN121997273ACN 121997273 ACN121997273 ACN 121997273ACN-121997273-A

Abstract

The invention relates to the technical field of dynamic management and control of lithium battery energy storage, and discloses a dynamic management and control method of a lithium battery energy storage system. The method comprises the steps of carrying out semantic analysis and enhancement on a text to obtain a state semantic vector, carrying out decomposition and feature synthesis on the waveform data to obtain an environment waveform vector, carrying out multi-level fusion on the environment waveform vector and the environment waveform vector to form a fusion feature map, constructing a correlation network between a load demand and a battery state by utilizing a correlation relationship in an analysis architecture mining map, carrying out collaborative analysis on the basis of the network to generate a dynamic management and control command for modifying identification parameters and optimizing strategies, and constructing an adjustment plan according to the dynamic management and control command. The method realizes the deep analysis and fusion of the multi-source heterogeneous data, and improves the accuracy and response adaptability of the dynamic management and control of the system.

Inventors

SUN QINRONG
JIANG YICHUN
DAI JINHONG
YIN LIMENG
ZHANG LONG
FENG WEI
WANG QIHUI

Assignees

重庆科技大学

Dates

Publication Date: 20260508
Application Date: 20260206

Claims (10)

1. The dynamic control method of the lithium battery energy storage system is characterized by realizing dynamic control through a series of sequentially executed operation phases, wherein the specific operation phases are as follows: Collecting an operation record data set of a target energy storage system in a continuous time interval, wherein the operation record data set consists of a battery state report text, environment monitoring waveform data and a load demand configuration file; Executing semantic analysis and enhancement operation on the battery state report text, outputting a state semantic vector, executing waveform decomposition and feature synthesis operation on the environment monitoring waveform data, outputting an environment waveform vector, and performing multi-level fusion operation on the state semantic vector and the environment waveform vector to generate a fusion feature map; Performing association relation mining operation on the fusion characteristic spectrum by adopting a preconfigured analysis framework to obtain an association relation network between load demand configuration and battery state report; based on the association relation network, executing cooperative analysis operation on the load demand configuration file and the battery state report text, and generating a dynamic management and control command, wherein the dynamic management and control command is used for identifying a modification item of a battery operation parameter and an optimization item of a load scheduling strategy; and constructing a parameter adjustment plan of the energy storage system according to the dynamic control command, and transmitting the parameter adjustment plan to a control interface of the energy storage system to start parameter change.
2. The method of claim 1, wherein the semantic parsing and enhancing operations are performed on the battery status report text, and the operation phase of outputting the status semantic vector involves the following processes: splitting the battery status report text into a plurality of semantic paragraph units; Executing semantic conversion processing of domain knowledge injection on each semantic paragraph unit to generate a basic semantic vector; Accessing a special term library of the lithium battery, extracting term groups related to the semantic paragraph units, and executing term encoding processing on the term groups to generate term vector groups; performing vector space alignment and merging processing on the basic semantic vector and the term vector group to obtain an enhanced semantic vector; The enhanced semantic vectors are concatenated into a state semantic vector in time stamp order.
3. The method of claim 1, wherein the operation of performing waveform decomposition and feature synthesis on the environmental monitoring waveform data, outputting the environmental waveform vector, involves: performing abnormal waveform segment positioning processing on the environment monitoring waveform data, and identifying abnormal waveform segments in the temperature waveform, the humidity waveform and the voltage waveform; Executing multi-resolution waveform slice processing in the abnormal waveform interval to obtain a plurality of waveform slices; Performing local waveform feature refinement processing on each waveform slice to generate a slice feature vector; All slice feature vectors are input into a waveform feature aggregator, the matching coefficient of each slice feature vector and a reference waveform template is calculated through a waveform pattern matching mechanism, weighted combination processing is carried out on the slice feature vectors based on the matching coefficient, and an environment waveform vector is output.
4. The method for dynamic management and control of a lithium battery energy storage system according to claim 1, wherein the operation phase of performing multi-level fusion operation on the state semantic vector and the environment waveform vector to generate the fusion characteristic map involves the following processes: Performing time axis synchronization processing on the state semantic vector and the environment waveform vector to enable the state semantic vector and the environment waveform vector to have consistent time scales; Establishing a cross attention mechanism between the semantic vector and the waveform vector, and calculating an interaction strength graph between the semantic component of each time scale in the state semantic vector and the waveform component of the corresponding time scale in the environment waveform vector; The interaction intensity diagram is utilized to guide the state semantic vector and the environment waveform vector to execute bidirectional feature tuning processing, and a tuned semantic vector and a tuned waveform vector are generated; and inputting the tuned semantic vector and the tuned waveform vector into a feature map generator, and outputting a fusion feature map through feature interleaving and dimension reduction operation.
5. The method for dynamic management and control of a lithium battery energy storage system according to claim 1, wherein the operation phase of performing an association mining operation on the fusion feature map by using a preconfigured analysis architecture to obtain an association network between load demand configuration and battery status report involves the following processes: inputting the fusion characteristic spectrum into a characteristic purification module of an analysis framework, removing irrelevant characteristics, and reserving a core characteristic spectrum; In a relation construction module of the analysis framework, resolving implicit relation between each feature dimension in the core feature map and a load demand configuration file and a battery state report text; Calculating the influence weight of each characteristic dimension on the relevance of load demand configuration and battery state report according to the implicit relation to form an influence weight list; And performing feature recombination and mapping processing on the core feature map based on the influence weight list to generate an association relation network, wherein the association relation network represents association paths between various parameters in load demand configuration and various indexes in a battery state report in a graph structure.
6. The method of claim 1, wherein the operation phase of generating the dynamic management command involves the following processes by performing collaborative analysis on the load demand profile and the battery status report text based on the association network: Traversing the association relation network, screening out association paths with association strength higher than a set value, and forming a key association set; aiming at each associated path in the key associated set, corresponding load demand deviation information and battery state abnormality information are decoded; The load demand deviation information and the battery state abnormality information are sent to a collaborative analyzer to be deduced, and a plurality of candidate management and control sequences are generated; And executing sequence evaluation and screening processing on each candidate control sequence, and selecting the optimal candidate control sequence as a dynamic control command.
7. The method of dynamic management of a lithium battery energy storage system according to claim 6, the load demand deviation information and the battery state abnormality information are sent to a collaborative analyzer to be deduced, and the operation stage of generating a plurality of candidate management and control sequences involves the following processes: performing deviation encoding processing on the load demand deviation information, and converting the load demand deviation information into a demand deviation characteristic representation; Performing exception coding processing on the battery state exception information, and converting the battery state exception information into state exception characteristic representation; Constructing an inference map between the demand deviation feature representation and the state anomaly feature representation, and performing iterative inference propagation on the inference map; And extracting an inference chain connecting the demand deviation characteristic representation and the state abnormal characteristic representation from the inference map, wherein each inference chain corresponds to one candidate management and control sequence.
8. The method of claim 6, wherein the operation phase of selecting the best-fit candidate control sequence as the dynamic control command involves the following steps: Analyzing the operation node chains in each candidate control sequence; invoking a sequence evaluation model to execute logic compliance inspection and resource demand measurement on the operation node chain; integrating the logic compliance checking result and the resource demand measuring and calculating result, and calculating the fitness value of each candidate control sequence; and selecting the candidate control sequence with the highest adaptation degree value as the optimal adaptation candidate control sequence.
9. The method of claim 1, wherein the operating phase of constructing the parameter adjustment plan of the energy storage system according to the dynamic management command involves the following processes: analyzing a battery operation parameter modification item and a load scheduling strategy optimization item in the dynamic management and control command; searching a history management log library, and searching a history execution record matched with a battery operation parameter modification item and a load scheduling strategy optimization item; filtering effective execution fragments from the historical execution records, and integrating the effective execution fragments into an initial adjustment scheme set; and executing multi-condition optimization processing on the initial adjustment scheme set, and outputting a parameter adjustment plan, wherein the multi-condition optimization processing comprises collaborative optimization of system stability constraint, energy consumption constraint and response time constraint.
10. The method of dynamic management of a lithium battery energy storage system of claim 1, further comprising: And updating the pre-configured analysis framework according to the system response data after the parameter adjustment plan is executed, wherein the updating operation comprises the steps of collecting operation monitoring data after parameter change, comparing actual operation characteristics with expected operation characteristics, and adjusting the internal parameter configuration of a characteristic purification module and a relation construction module in the analysis framework according to the comparison difference.

Description

Dynamic control method of lithium battery energy storage system Technical Field The invention relates to the technical field of dynamic management and control of lithium battery energy storage, in particular to a dynamic management and control method of a lithium battery energy storage system. Background In the current dynamic management and control method of the lithium battery energy storage system, a separate processing mode is often adopted at the data processing level. For text data such as battery status reports, the prior art relies on keyword matching or simple classification for status recognition, and lacks deep semantic understanding. For environmental monitoring waveform data, conventional methods commonly employ direct statistical feature extraction or fixed-threshold waveform comparison, simplifying complex continuous waveforms into few statistics. In the feature fusion stage, the conventional scheme generally performs simple vector splicing or shallow model-based fusion on the processed text features and waveform features, and fails to fully mine deep and nonlinear interaction relations among heterogeneous data. The above-described separated and coarse-grained data processing scheme has drawbacks. The lack of text semantic understanding results in failure to accurately capture the implicit complex working condition description in the state report, and the simple statistical processing of waveforms loses the dynamic details and frequency domain characteristics of the waveforms which change along with time, so that the recognition capability of sudden environmental disturbance or gradual change failure modes is insufficient. However, the feature fusion of the shallow layer is difficult to establish a precise and interpretable association model between the battery state, the environmental fluctuation and the load demand, so that the generation of a system management strategy depends on experience and rough mapping, and precise dynamic regulation and control based on multidimensional depth association cannot be realized. How to analyze the environment waveform more finely so as to keep key dynamic information and how to realize depth and structuring fusion between the text and the waveform heterogeneous characteristics is a key problem for improving the dynamic management and control precision of the energy storage system. Disclosure of Invention The invention aims to provide a dynamic control method of a lithium battery energy storage system, which aims to solve the problems in the background technology. In order to achieve the above object, the present invention provides a method for dynamically controlling an energy storage system of a lithium battery, the method comprising: Collecting an operation record data set of a target energy storage system in a continuous time interval, wherein the operation record data set consists of a battery state report text, environment monitoring waveform data and a load demand configuration file; Executing semantic analysis and enhancement operation on the battery state report text, outputting a state semantic vector, executing waveform decomposition and feature synthesis operation on the environment monitoring waveform data, outputting an environment waveform vector, and performing multi-level fusion operation on the state semantic vector and the environment waveform vector to generate a fusion feature map; Performing association relation mining operation on the fusion characteristic spectrum by adopting a preconfigured analysis framework to obtain an association relation network between load demand configuration and battery state report; based on the association relation network, executing cooperative analysis operation on the load demand configuration file and the battery state report text, and generating a dynamic management and control command, wherein the dynamic management and control command is used for identifying a modification item of a battery operation parameter and an optimization item of a load scheduling strategy; and constructing a parameter adjustment plan of the energy storage system according to the dynamic control command, and transmitting the parameter adjustment plan to a control interface of the energy storage system to start parameter change. Preferably, the semantic parsing and enhancing operation is performed on the battery state report text, and the operation stage of outputting the state semantic vector involves the following processes: splitting the battery status report text into a plurality of semantic paragraph units; Executing semantic conversion processing of domain knowledge injection on each semantic paragraph unit to generate a basic semantic vector; Accessing a special term library of the lithium battery, extracting term groups related to the semantic paragraph units, and executing term encoding processing on the term groups to generate term vector groups; performing vector space alignment and merging processing on the bas