Extreme environment energy storage regulation and control method and system based on interconnected large model

Abstract

The invention relates to the technical field of large models, and discloses an extreme environment energy storage regulation and control method and system based on an interconnected large model, and the method comprises the steps: simulating a federated learning environment of energy storage equipment in an extreme environment, and constructing a global energy storage regulation and control large model of the energy storage equipment; collecting a multi-modal time sequence data set of the energy storage equipment in an extreme environment to construct an environment perception vector of the energy storage equipment; the extreme environment event probability and the equipment operation risk of the energy storage equipment are analyzed; the method comprises the steps of generating a regulation and control instruction of the energy storage equipment, simulating the regulation and control instruction to obtain a simulation result, and analyzing the performability and safety of the regulation and control instruction based on the simulation result to finally generate a decision instruction of the energy storage equipment; and adjusting operation parameters of the energy storage equipment to execute regulation and control of the energy storage equipment in the extreme environment. Accuracy and adaptability of energy storage regulation and control in the extreme environment can be improved.

Description

Technical Field

[0001] This invention relates to a method and system for extreme environment energy storage regulation based on an interconnected large model, belonging to the field of large model technology. Background Technology

[0002] Extreme environment energy storage regulation refers to the dynamic, intelligent, and precise adjustment and control of energy storage systems operating under extreme environmental conditions through advanced control strategies and technologies. This ensures that the system can still operate safely, efficiently, and reliably under harsh conditions and extends its service life.

[0003] Traditional extreme environment energy storage regulation methods employ fixed control strategies, such as those described in CN117060476B. These strategies involve setting upper and lower thresholds for key parameters like temperature, voltage, and current, triggering protective actions (such as shutdown or reduced power operation) when real-time data exceeds these thresholds. These methods rely on data from a single sensor, making it difficult to comprehensively perceive complex and ever-changing extreme environments. Data utilization is limited to local devices, creating "data silos" that cannot be shared or learned across devices. The models have poor generalization capabilities, making them ill-suited for unknown and complex extreme conditions. Furthermore, the lack of simulation verification mechanisms for regulation commands introduces execution risks. Therefore, traditional methods have significant shortcomings in terms of dynamic environment adaptability, global optimization capabilities, and safety. Summary of the Invention

[0004] This invention provides a method and system for extreme environment energy storage regulation based on an interconnected large model, the main purpose of which is to improve the accuracy and adaptability of extreme environment energy storage regulation.

[0005] To achieve the above objectives, this invention provides an extreme environment energy storage regulation method based on an interconnected large-scale model, comprising: The extreme environmental characteristics of the energy storage device are obtained to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, a global energy storage regulation model of the energy storage device is constructed. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer. The sensing layer collects multimodal time-series datasets of energy storage devices under extreme environments, including real-time meteorological data, equipment operating status data, and historical fault database data, to construct the environmental perception vector of the energy storage device. Based on the environmental perception vector, the analysis layer analyzes the probability of extreme environmental events and the operational risks of the energy storage device. Based on the probability of extreme environmental events and the operational risks of the equipment, the decision-making layer initially generates control commands for the energy storage device, simulates the control commands, obtains simulation results, and analyzes the executability and security of the control commands based on the simulation results, so as to finally generate decision commands for the energy storage device. Based on the decision instructions, the operating parameters of the energy storage device are adjusted through the execution layer to perform regulation of the energy storage device under the extreme environment.

[0006] Optionally, the step of constructing a global energy storage regulation model for the energy storage device based on the federated learning environment includes: Configure the sensors of the energy storage device to construct a sensor network for the energy storage device; Based on the sensor network, the sensing layer of the energy storage device is determined; Based on the federated learning environment, a global environment analysis federated model of the energy storage device is constructed to determine the analysis layer of the energy storage device. Define the control objectives of the energy storage device in order to construct the target excitation function of the energy storage device; Based on the target incentive function, the decision layer of the energy storage device is constructed; Determine the PLC algorithm of the energy storage device to construct the execution layer of the energy storage device; By integrating the perception layer, the analysis layer, the decision-making layer, and the execution layer, a global energy storage regulation model is obtained.

[0007] Optionally, the step of constructing a global environment analysis federated model for the energy storage device based on the federated learning environment includes: Obtain the local dataset corresponding to the local node in the federated learning environment; The global environment analysis model corresponding to the federated learning environment is deployed to the local node so as to train the global environment analysis model using the local dataset to obtain the local environment analysis model. Extract the model parameters of the local environment analysis model; The model parameters are aggregated to obtain aggregated parameters; The global environment analysis model is updated using the aggregation parameters to obtain a global environment analysis federated model. Optionally, the objective excitation function for constructing the energy storage device includes: The target indicators corresponding to the regulation target of the energy storage device are quantified, wherein the target indicators include: safety indicators, economic indicators, stability indicators and lifespan indicators; Based on the target indicators, a plurality of incentive functions are constructed for the energy storage device, wherein the plurality of incentive functions include: a safety incentive function, an economic incentive function, a stability incentive function, and a lifetime incentive function; Determine the regulation sensitivity of the energy storage device in order to calculate the dynamic multinomial weights of the multinomial excitation functions; Calculate the extreme environmental factors of the energy storage device under the extreme environment; The target activation function is obtained by weighting the multinomial activation functions based on the dynamic multinomial weights and the extreme environmental factors.

[0008] Optionally, the federated learning environment simulating energy storage devices under extreme environments includes: Determine the federated learning framework for the energy storage device; Determine the central server of the federated learning framework and the local nodes of the energy storage device; Establish the communication protocol and encryption algorithm between the central server and the local nodes; Based on the extreme environmental characteristics corresponding to the energy storage device, a global environment analysis model of the federated learning framework is defined. The federated learning environment of the energy storage device is determined based on the communication protocol, the encryption algorithm, and the global environment analysis model.

[0009] Optionally, constructing the environmental perception vector of the energy storage device includes: The multimodal time series dataset corresponding to the energy storage device is preprocessed to obtain a preprocessed multimodal time series dataset; Extract multimodal data features from the preprocessed multimodal time-series dataset; The multimodal data features are fused to obtain high-dimensional fused features; The high-dimensional fusion features are reduced in dimensionality to obtain the dimensionality-reduced fusion features; Based on the aforementioned dimensionality reduction and fusion features, an environmental perception vector for the energy storage device is constructed.

[0010] Optionally, the step of analyzing the probability of extreme environmental events and the operational risks of the energy storage device through the analysis layer based on the environmental perception vector includes: Extract the meteorological feature sequence and equipment operating status sequence from the environmental perception vector; Calculate the probability of extreme environmental events for the energy storage device based on the meteorological characteristic sequence; Based on the probability of the extreme environmental events, identify the operational risk indicators of the energy storage device; The operational risk of the energy storage device is calculated based on the operational status sequence and the operational risk indicators.

[0011] Optionally, the step of generating preliminary control instructions for the energy storage device through the decision-making layer based on the probability of the extreme environmental event and the operational risk of the device includes: Calculate the comprehensive risk index of the energy storage device based on the probability of the extreme environmental events and the operational risks of the device. Based on the comprehensive risk index, the potential risk factors of the energy storage device are determined; Based on the potential risk factors, the instruction generation logic of the energy storage device is determined; Based on the instruction generation logic, the control instructions for the energy storage device are initially generated.

[0012] Optionally, the analysis of the executability and security of the control command includes: The simulation results corresponding to the control commands are normalized to obtain normalized simulation results; Extract the equipment state characteristics, system response characteristics, and environmental parameters from the normalized simulation results; Based on the device status characteristics, determine the simulated operating status of the energy storage device corresponding to the control command; Calculate the safety coefficient of the simulation running state to determine the safety of the control commands; Based on the system response characteristics, analyze the timing feasibility of the control command; Based on the environmental parameters, the environmental adaptability of the control command is determined; The executability of the control command is determined by considering the aforementioned security, time feasibility, and environmental adaptability.

[0013] To address the aforementioned problems, this invention also provides an extreme environment energy storage and regulation system based on an interconnected large-scale model, the system comprising: The large-scale regulation model construction module is used to acquire the extreme environmental characteristics of the energy storage device to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, a global energy storage regulation model of the energy storage device is constructed. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer. An environmental perception module is used to collect multimodal time-series datasets of energy storage devices under extreme environments through the perception layer. The multimodal time-series datasets include real-time meteorological data, equipment operating status data, and historical fault database data to construct the environmental perception vector of the energy storage device. The risk analysis module is used to analyze the probability of extreme environmental events and the operational risks of the energy storage device through the analysis layer based on the environmental perception vector. The instruction generation module is used to generate preliminary control instructions for the energy storage device through the decision layer based on the probability of the extreme environmental events and the operational risks of the device, simulate the control instructions to obtain simulation results, and analyze the executability and security of the control instructions based on the simulation results, so as to finally generate decision instructions for the energy storage device. The energy storage regulation module is used to adjust the operating parameters of the energy storage device through the execution layer based on the decision command, so as to perform regulation of the energy storage device in the extreme environment.

[0014] Compared to the problems described in the background technology, this invention constructs a global energy storage regulation model using federated learning technology. This model effectively integrates data resources from multiple sources, improving the model's generalization ability and privacy protection. The model comprises a perception layer, an analysis layer, a decision-making layer, and an execution layer, forming a complete closed-loop intelligent regulation system. The perception layer collects real-time meteorological data, equipment operating status data, and historical fault database data to construct a multimodal time-series dataset, forming an environmental perception vector that comprehensively characterizes the external environment and internal state of the energy storage equipment. Based on this vector, the analysis layer accurately predicts the probability of extreme environmental events and equipment operating risks, providing a scientific basis for decision-making. The decision-making layer generates preliminary regulation commands, evaluates their feasibility and safety through simulation, and ultimately outputs reliable decision commands, ensuring the scientific rationality of the regulation strategy. The execution layer dynamically adjusts equipment operating parameters according to the decision commands, achieving precise regulation of energy storage equipment in extreme environments. This invention not only improves the adaptability and safety of energy storage systems in complex environments but also significantly optimizes energy dispatch efficiency and equipment lifespan, providing key technical support for building an intelligent and resilient energy internet and promoting the intelligent and sustainable development of energy systems. Therefore, the extreme environment energy storage regulation method based on the interconnected large model provided in this embodiment of the invention can improve the accuracy and adaptability of extreme environment energy storage regulation. Attached Figure Description

[0015] Figure 1 A flowchart illustrating an extreme environment energy storage regulation method based on an interconnected large model, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the model construction process for an extreme environment energy storage and control system based on an interconnected large model, provided as an embodiment of the present invention.

[0016] Figure 3 This is a schematic diagram of a module for implementing the extreme environment energy storage and control system based on the interconnected large model, provided as an embodiment of the present invention.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for extreme environment energy storage regulation based on an interconnected big data model. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating an extreme environment energy storage regulation method based on an interconnected large-scale model according to an embodiment of the present invention. In this embodiment, the extreme environment energy storage regulation method based on an interconnected large-scale model includes: S1. Obtain the extreme environmental characteristics of the energy storage device to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, construct a global energy storage regulation model for the energy storage device. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer.

[0021] This invention provides an environmental data foundation for subsequent analysis by acquiring the extreme environmental characteristics of energy storage devices. These extreme environmental characteristics refer to the quantitative representation of internal and external environmental factors that significantly deviate from the normal operating conditions of the energy storage device and may have a serious negative impact on its safety, reliability, lifespan, and performance, such as temperature characteristics, atmospheric pressure characteristics, and wind characteristics.

[0022] This invention, through a federated learning environment simulating energy storage devices under extreme conditions, can simulate hundreds or thousands of virtual energy storage nodes in different extreme environments in parallel, thereby significantly shortening the algorithm convergence time and accelerating the iteration and optimization process. The federated learning environment refers to a technical framework and infrastructure customized for energy storage devices, enabling collaborative machine learning under simulated extreme conditions.

[0023] As an embodiment of the present invention, the federated learning environment simulating energy storage devices under extreme environments includes: Determine the federated learning framework for the energy storage device; Determine the central server of the federated learning framework and the local nodes of the energy storage device; Establish the communication protocol and encryption algorithm between the central server and the local nodes; Based on the extreme environmental characteristics corresponding to the energy storage device, a global environment analysis model of the federated learning framework is defined. The federated learning environment of the energy storage device is determined based on the communication protocol, the encryption algorithm, and the global environment analysis model.

[0024] The federated learning framework refers to a decentralized collaborative machine learning system architecture specifically designed for distributed energy storage device networks, such as Flower (FLwr) and FedML. The central server, within the federated learning framework, is the core computing and control node responsible for coordinating local model training across various energy storage devices (clients), aggregating model parameters, generating a global model, performing macro-level analysis and decision-making based on the global model, and ultimately issuing control commands. Local nodes, within the federated learning framework, are edge computing units deployed on individual energy storage devices, responsible for collecting local operating and environmental data, performing local model training, uploading model updates, and receiving and executing global control commands. The communication protocol refers to a set of standardized rules and formats followed for data exchange between the central server and local nodes, such as Protocol Buffers (Protobuf) and MessagePac. The encryption algorithm refers to a series of cryptographic techniques used to protect the privacy and security of model parameters uploaded by local nodes during transmission and aggregation, such as homomorphic encryption algorithms and symmetric encryption algorithms. The global environment analysis model refers to a machine learning model that is collaboratively trained by all local nodes in a federated learning framework, ultimately shared, and used to learn and quantify the complex mapping relationship between extreme environment characteristics and the operating status of energy storage devices.

[0025] Optionally, the federated learning framework can be determined through framework comparison tests in scenario simulations, such as simulating the performance of the energy storage device under extreme conditions such as network jitter, node failure, and model poisoning attacks, in order to determine the federated learning framework of the energy storage device.

[0026] Optionally, the communication protocol can be constructed using reinforcement learning techniques, such as dynamically adjusting communication strategies (including node selection, communication frequency, and data compression rate) to adapt to uncertainties such as bandwidth fluctuations and node failures in extreme environments. The encryption algorithm can be constructed using homomorphic encryption schemes that support floating-point operations, such as CKKS or BFV.

[0027] This invention, through the federated learning environment, constructs a global energy storage control model for the energy storage device, enabling it to perceive and adapt to extreme environments (such as high temperatures, extreme cold, typhoons, earthquakes, etc.), thus ensuring the safe operation of the energy storage device under harsh conditions. The global energy storage control model refers to an intelligent, collaborative, and adaptive energy control model constructed based on a federated learning environment, composed of multiple geographically distributed but logically unified energy storage devices and their control systems.

[0028] As an embodiment of the present invention, the step of constructing a global energy storage regulation model for the energy storage device based on the federated learning environment includes: Configure the sensors of the energy storage device to construct a sensor network for the energy storage device; Based on the sensor network, the sensing layer of the energy storage device is determined; Based on the federated learning environment, a global environment analysis federated model of the energy storage device is constructed to determine the analysis layer of the energy storage device. Define the control objectives of the energy storage device in order to construct the target excitation function of the energy storage device; Based on the target incentive function, the decision layer of the energy storage device is constructed; Determine the PLC algorithm of the energy storage device to construct the execution layer of the energy storage device; By integrating the perception layer, the analysis layer, the decision-making layer, and the execution layer, a global energy storage regulation model is obtained.

[0029] The sensors mentioned here refer to physical devices used for real-time monitoring of the operating status and environmental parameters of energy storage equipment, such as temperature sensors, voltage sensors, and current sensors. The sensor network refers to a distributed monitoring network composed of multiple sensors. The perception layer refers to a data acquisition and status monitoring layer built upon the sensor network. The global environmental analysis federated model refers to a distributed machine learning model built upon a federated learning framework. The analysis layer refers to a data processing and intelligent analysis layer built upon the global environmental analysis federated model, responsible for in-depth analysis, feature extraction, pattern recognition, and trend prediction of the data from the perception layer, providing data support for the decision-making layer. The control objectives refer to the key performance indicators that need to be optimized during the operation of the energy storage equipment, including multiple dimensions such as safety, economy, stability, and lifespan. The target incentive function refers to a mathematical function that quantifies and weights multiple control objectives. The decision-making layer refers to an intelligent decision-making layer that generates the optimal control strategy through reinforcement learning, optimization algorithms, etc. The PLC algorithm refers to a control logic algorithm running on a programmable logic controller (PLC). The execution layer refers to a device control layer built upon the PLC algorithm, responsible for translating decision commands into physical operations, achieving precise control of the energy storage equipment, and feeding back the execution results to the perception layer to form a closed loop.

[0030] Optionally, the sensor network can be constructed using fiber optic sensing and distributed acoustic sensing, such as using distributed acoustic sensing and distributed temperature sensing technologies to monitor physical quantities such as temperature, strain, and vibration of the energy storage device in real time.

[0031] Optionally, the PLC algorithm can be determined using reinforcement learning techniques. For example, by continuously interacting with the environment, reinforcement learning can autonomously learn the optimal control strategy to achieve dynamic optimization of the PLC algorithm.

[0032] Optionally, the step of constructing a global environment analysis federated model for the energy storage device based on the federated learning environment includes: Obtain the local dataset corresponding to the local node in the federated learning environment; The global environment analysis model corresponding to the federated learning environment is deployed to the local node so as to train the global environment analysis model using the local dataset to obtain the local environment analysis model. Extract the model parameters of the local environment analysis model; The model parameters are aggregated to obtain aggregated parameters; The global environment analysis model is updated using the aggregation parameters to obtain a global environment analysis federated model. The local dataset refers to the raw data owned by each local node (such as an energy storage device, edge gateway, or regional controller) deployed in the federated learning environment, which is not shared with other nodes. The local environment analysis model refers to a personalized model tailored to the specific environmental characteristics of a local node, obtained by training the global environment analysis model on the local dataset after deployment to the local node. The model parameters refer to the transferable numerical information such as weights, biases, and structural configurations learned after the local environment analysis model is trained. The aggregation parameters refer to the globally representative parameters calculated by the central server after collecting model parameters from multiple local nodes.

[0033] To further understand the global environment analysis federated model, please refer to [link / reference]. Figure 2 The diagram shown illustrates the model construction process of an extreme environment energy storage and control system based on an interconnected large model, according to an embodiment of the present invention. Figure 2 The process begins by acquiring a local dataset, using it to train a global model to obtain a local environment analysis model, and then extracting the parameters of the local model and uploading them to the server. Upon receiving the parameters, the server first determines whether parameter aggregation across all nodes is complete. If complete, it updates the global model with the aggregated parameters, generating a new global environment analysis federated model, and then determines whether to continue to the next round of training. If not, it outputs the final model.

[0034] Optionally, the aggregation parameters can be obtained by an aggregation algorithm, such as FedAvg or FedProx. The aggregation parameters are obtained by using the FedAvg algorithm to perform a weighted average of the model parameters of each local node.

[0035] Optionally, the objective excitation function for constructing the energy storage device includes: The target indicators corresponding to the regulation target of the energy storage device are quantified, wherein the target indicators include: safety indicators, economic indicators, stability indicators and lifespan indicators; Based on the target indicators, a plurality of incentive functions are constructed for the energy storage device, wherein the plurality of incentive functions include: a safety incentive function, an economic incentive function, a stability incentive function, and a lifetime incentive function; Determine the regulation sensitivity of the energy storage device in order to calculate the dynamic multinomial weights of the multinomial excitation functions; Calculate the extreme environmental factors of the energy storage device under the extreme environment; The target activation function is obtained by weighting the multinomial activation functions based on the dynamic multinomial weights and the extreme environmental factors.

[0036] The target indicators refer to key performance parameters used to quantify the control effect of energy storage devices. The safety indicators refer to values ​​reflecting whether the equipment operates within a safe range, such as temperature safety thresholds and current safety thresholds. The economic indicators refer to values ​​reflecting the economic benefits of equipment operation, such as electricity price differences and charging / discharging power. The stability indicators refer to values ​​reflecting the equipment's ability to support the grid and load, such as frequency deviation, voltage deviation, and response time. The lifespan indicators refer to values ​​reflecting the degree of equipment aging and remaining lifespan, such as depth of charge / discharge and number of charge / discharge cycles. The multinomial incentive functions are mathematical functions constructed based on the target indicators to quantitatively evaluate the control behavior of energy storage devices. The safety incentive function is a function used to quantify whether the equipment's operating state is within a safe range, usually in the form of a penalty function; the closer to the safety boundary, the lower the incentive value. The economic incentive function is a function used to quantify the economic benefits of equipment operation, usually adopting the principle of maximizing benefits or minimizing costs. The stability incentive function is a function used to quantify the equipment's ability to support the grid and load, usually with the objective of minimizing deviation. The lifespan incentive function is a function used to quantify the degree of equipment aging, usually with the objective of minimizing cycle life decay. The regulation sensitivity refers to the degree of response of the energy storage device to different incentive targets, reflecting the device's sensitivity to targets such as safety, economy, stability, and lifespan during regulation. The dynamic multi-weight refers to the weight coefficients of various incentive functions calculated in real time based on regulation sensitivity, current operating status, and environmental factors. The extreme environmental factor refers to parameters used to quantify the severity of the current environment, reflecting the impact of extreme conditions such as temperature, humidity, wind speed, and sudden load changes on device operation.

[0037] Optionally, the regulation sensitivity of the energy storage device can be determined by deep learning technology, such as using deep neural networks or long short-term memory networks to model the historical operating data of the energy storage device, and quantifying the sensitivity of different regulation parameters (such as charging and discharging power, temperature, voltage, etc.) to target indicators (such as safety, economy, stability, lifespan) through gradient analysis or attention mechanisms.

[0038] Optionally, the extreme environmental factors can be calculated using multiphysics coupling simulation technology, such as using finite element analysis and computational fluid dynamics to establish a thermo-electric-mechanical-chemical multi-field coupling model of the energy storage device under extreme environments, and then calculating the environmental factors through simulation.

[0039] As another implementation, the polynomial excitation function can be expressed by the following formula:

[0040]

[0041]

[0042]

[0043] in, This represents the safety value of the safety excitation function. This represents the economic value of the economic incentive function. This represents the stable value of the stable excitation function. This represents the lifetime value of the lifetime excitation function. Indicates the energy storage device corresponding to the first Safety indicators Indicates the energy storage device corresponding to the first The safety threshold of each safety indicator This indicates the total number of safety indicators. Indicates the economic efficiency coefficient. The charge-discharge price difference represents an economic indicator. The charging and discharging power represents an economic indicator. This represents the error stability coefficient. Represents the response stability coefficient. The frequency deviation of the stability index Voltage deviation, representing a stability indicator. The response time of the stability index. Indicates the depth of charge / discharge factor. Indicates the depth of charge and discharge. Indicates the charge / discharge cycle coefficient. Indicates the number of charge / discharge cycles.

[0044] What needs to be explained is that in this application, in the formula... The economic efficiency coefficient is used to balance economic benefits and risks. It can be determined through regression analysis of historical operating data and takes a value of [0.5, 2]. The larger the value, the higher the economic efficiency weight. For example, the typical value of 1.2 can be used in this application. This represents the error stability coefficient, used to quantify the contribution to grid stability. It can be determined through grid sensitivity analysis, and its value ranges from [0.01, 0.1]. The larger the value, the heavier the penalty for deviation. The initial value is usually set at 0.05. The response stability coefficient is used to quantify the impact of response speed. It can be calibrated through power grid stability simulation experiments and has a value range of [0.05, 0.2]. The larger the value, the heavier the penalty for slow response. For example, the value in this application is 0.1. The charge / discharge depth coefficient is used to quantify the impact of charge / discharge depth on battery life. It can be determined by a battery life model, such as the rainflow counting method combined with the Arrhenius aging model. The value range is [0.1, 0.5]. The larger the value, the heavier the penalty on charge / discharge depth. For example, the value in this application is 0.2. The charge / discharge cycle coefficient is used to quantify the impact of charge / discharge cycles on lifetime. It can be determined by the cycle lifetime curve and Miner's linear cumulative damage rule. The value range is [0.01, 0.1]. The larger the value, the heavier the penalty for high-frequency charge / discharge. For example, the value in this application is 0.05.

[0045] S2. Collect multimodal time-series datasets of energy storage devices under extreme environments through the perception layer, wherein the multimodal time-series datasets include: real-time meteorological data, equipment operating status data and historical fault database data, in order to construct the environmental perception vector of the energy storage device.

[0046] This invention, through the perception layer, collects multimodal time-series datasets of energy storage devices under extreme environments. This allows for comprehensive perception of the impact and operational data of energy storage devices in extreme environments, providing strong data support for intelligent control and global optimization. The multimodal time-series dataset refers to a collection of data from multiple sources and of various types, all organized in time-series format. Real-time meteorological data refers to a collection of data reflecting current or near-real-time environmental meteorological conditions, such as temperature, humidity, wind speed, air pressure, rainfall, and light intensity, acquired through various meteorological sensors using continuous or high-frequency sampling. Device operational status data refers to a real-time data collection reflecting the current operational status, performance parameters, health status, and operating environment of the energy storage device, collected through its built-in sensors, monitoring modules, or control systems. This includes data such as electrical parameters, thermal management parameters, mechanical status, and control status. Historical fault database data refers to a structured and semi-structured data collection accumulated during the long-term operation of the energy storage device, recording equipment fault events, causes, characteristics, handling measures, and subsequent impacts, such as fault codes, timestamps, and device IDs.

[0047] This invention, through the construction of an environmental perception vector for the energy storage device, can comprehensively characterize the external environment and internal state of the energy storage device, providing high-dimensional and high-precision environmental state input for subsequent intelligent regulation, fault prediction, and health management. Specifically, the environmental perception vector refers to a high-dimensional, structured vector representation that quantifies the external environmental state, internal operating state, and historical experience knowledge of the energy storage device through multimodal data fusion and feature extraction techniques.

[0048] As an embodiment of the present invention, constructing the environmental perception vector of the energy storage device includes: The multimodal time series dataset corresponding to the energy storage device is preprocessed to obtain a preprocessed multimodal time series dataset; Extract multimodal data features from the preprocessed multimodal time-series dataset; The multimodal data features are fused to obtain high-dimensional fused features; The high-dimensional fusion features are reduced in dimensionality to obtain the dimensionality-reduced fusion features; Based on the aforementioned dimensionality reduction and fusion features, an environmental perception vector for the energy storage device is constructed.

[0049] The preprocessed multimodal time-series dataset refers to the process of cleaning, aligning, standardizing, and enhancing the original collected multimodal time-series data (including real-time meteorological data, equipment operating status data, and historical fault database data) to obtain a high-quality, structured dataset. The multimodal data features refer to the effective features extracted from the preprocessed multimodal time-series dataset, including statistical features, frequency domain features, temporal domain features, and semantic features. The high-dimensional fusion feature refers to fusing the multimodal data features (meteorology, equipment, faults) into a high-dimensional vector, forming a unified representation of the environment and equipment status. The dimensionality-reducing fusion feature refers to compressing the high-dimensional fusion feature into a low-dimensional vector, removing redundant information, and retaining key features.

[0050] Optionally, the high-dimensional fusion features can be obtained through feature fusion algorithms, such as feature concatenation, weighted fusion, deep learning fusion, etc.

[0051] Optionally, the dimensionality reduction fusion features can be obtained through dimensionality reduction techniques, such as principal component analysis or linear discriminant analysis.

[0052] S3. Based on the environmental perception vector, the analysis layer analyzes the probability of extreme environmental events and the operational risks of the energy storage device.

[0053] This invention, through the analysis layer based on the environmental perception vector, analyzes the probability of extreme environmental events and the operational risks of the energy storage device, enabling proactive early warning and laying the foundation for subsequent energy storage regulation. The probability of extreme environmental events refers to the probability value of extreme events occurring in the external environment (such as meteorology, geography, power grid, etc.) that may significantly adversely affect the operational safety, performance, or lifespan of the energy storage device. The operational risk refers to the comprehensive risk level of the energy storage device due to factors such as internal state, external environment, or operating strategies, resulting in performance degradation, malfunctions, safety accidents, or even shutdowns.

[0054] As an embodiment of the present invention, the step of analyzing the probability of extreme environmental events and the operational risks of the energy storage device through the analysis layer based on the environmental perception vector includes: Extract the meteorological feature sequence and equipment operating status sequence from the environmental perception vector; Calculate the probability of extreme environmental events for the energy storage device based on the meteorological characteristic sequence; Based on the probability of the extreme environmental events, identify the operational risk indicators of the energy storage device; The operational risk of the energy storage device is calculated based on the operational status sequence and the operational risk indicators.

[0055] The meteorological feature sequence refers to multi-dimensional time-series data extracted from the environmental perception vector that is related to the external meteorological environment of the energy storage device, such as temperature, wind speed, rainfall, and thunderstorms. The device operating status sequence refers to multi-dimensional time-series data extracted from the environmental perception vector that reflects the current operating status of the energy storage device, such as battery temperature, battery voltage, and internal resistance. The operating risk index refers to quantitative indicators derived from the probability of extreme environmental events, used to assess the level of risk that the device may face under specific environmental conditions, such as thermal runaway risk, lightning strike risk, and heat dissipation failure risk.

[0056] Optionally, the probability of extreme environmental events of the energy storage device can be calculated using a statistical probability model, such as a generalized extreme value distribution or a generalized Pareto distribution.

[0057] Optionally, the equipment operation risk of the energy storage device can be calculated using a Bayesian network, such as by using a Bayesian network to model the causal relationship between the operating state, risk indicators and equipment risk, so as to calculate the equipment operation risk through conditional probability.

[0058] S4. Based on the probability of the extreme environmental events and the operational risks of the equipment, the decision-making layer initially generates control commands for the energy storage device, simulates the control commands, obtains simulation results, and analyzes the executability and security of the control commands based on the simulation results, so as to finally generate decision commands for the energy storage device.

[0059] This invention, through its embodiments, uses the decision-making layer to initially generate control commands for the energy storage device based on the probability of extreme environmental events and the operational risks of the equipment. This allows for the early identification and mitigation of potential risks, reducing equipment failures and safety accidents, dynamically adjusting charging and discharging strategies, lowering operating costs, and increasing profitability. Specifically, the control commands refer to the detailed operational instructions generated by the decision-making layer during the operation of the energy storage device, based on the probability of extreme environmental events and the operational risks, to control the device's operating status, parameters, or strategies.

[0060] As an embodiment of the present invention, the step of generating preliminary control instructions for the energy storage device through the decision-making layer based on the probability of the extreme environmental event and the operational risk of the device includes: Calculate the comprehensive risk index of the energy storage device based on the probability of the extreme environmental events and the operational risks of the device. Based on the comprehensive risk index, the potential risk factors of the energy storage device are determined; Based on the potential risk factors, the instruction generation logic of the energy storage device is determined; Based on the instruction generation logic, the control instructions for the energy storage device are initially generated.

[0061] The comprehensive risk index is a quantitative indicator calculated based on the probability of extreme environmental events and equipment operational risks. The potential risk factors are specific risk sources identified that have a significant impact on the overall risk of energy storage equipment. The instruction generation logic is a decision-making mechanism that transforms risk assessment results into specific control instructions based on potential risk factors and the comprehensive risk index.

[0062] Optionally, the comprehensive risk index of the energy storage device can be calculated using the fuzzy comprehensive evaluation method, such as dividing environmental risk and equipment risk into multiple levels, and calculating the comprehensive risk index through membership functions and fuzzy rules.

[0063] Optionally, the potential risk factors can be determined through risk source analysis, such as analyzing the main contributing sources of the comprehensive risk index to identify potential risk factors of energy storage equipment.

[0064] Optionally, the instruction generation logic of the energy storage device can be implemented using deep reinforcement learning algorithms (such as DQN, PPO), and the optimal control instructions can be generated by training in a simulation environment through Q-learning networks or policy gradient methods.

[0065] This invention, through simulation of the control commands, verifies the rationality, safety, and effectiveness of the commands, thereby avoiding equipment damage, system instability, or economic losses caused by improper commands. The simulation results refer to various data, indicators, and visualizations reflecting the operating status and performance of the energy storage device under simulated conditions, output after simulating the control commands.

[0066] Optionally, the simulation results can be obtained using a digital twin simulation method.

[0067] This invention, through analyzing the executability and safety of control commands, can identify and resolve problems in the commands in advance, significantly improving the reliability of decision-making and avoiding equipment damage, system failures, and safety accidents caused by improper commands. Executability refers to the ability of a control command to be correctly, completely, and safely executed by equipment in a real physical environment. Safety refers to the ability of the control command to not cause damage to equipment, personnel, power grid, or the environment during execution.

[0068] As an embodiment of the present invention, the analysis of the executability and security of the control command includes: The simulation results corresponding to the control commands are normalized to obtain normalized simulation results; Extract the equipment state characteristics, system response characteristics, and environmental parameters from the normalized simulation results; Based on the device status characteristics, determine the simulated operating status of the energy storage device corresponding to the control command; Calculate the safety coefficient of the simulation running state to determine the safety of the control commands; Based on the system response characteristics, analyze the timing feasibility of the control command; Based on the environmental parameters, the environmental adaptability of the control command is determined; The executability of the control command is determined by considering the aforementioned security, time feasibility, and environmental adaptability.

[0069] The normalized simulation results refer to the original simulation data processed using standardization methods to map it to a specific interval (e.g., [0, 1] or [-1, 1]) to eliminate dimensional differences and facilitate subsequent feature extraction, model input, and comprehensive comparison. The equipment state characteristics refer to key parameters extracted from the normalized simulation results that reflect the internal operating state of the energy storage device during the execution of control commands. The system response characteristics refer to parameters extracted from the normalized simulation results that reflect the dynamic characteristics of the energy storage system's response to control commands. The environmental parameters refer to parameters extracted from the normalized simulation results that reflect the external environmental conditions of the energy storage device. The simulated operating state refers to the comprehensive judgment of the simulated operating mode of the energy storage device during the execution of control commands based on the equipment state characteristics, such as normal operating state, overload operating state, and abnormal state. The safety coefficient is an indicator used to comprehensively evaluate the safety risk level of the energy storage device during the execution of control commands. The time feasibility refers to the ability of the system to effectively execute control commands within a specified time. The environmental adaptability refers to the ability of the control commands to be executed safely and stably under specific environmental conditions.

[0070] Optionally, the normalized simulation results can be obtained through standardization methods, such as Min-Max, Z-score, decimal scaling, etc.

[0071] Optionally, the safety coefficient can be calculated by fuzzy comprehensive evaluation, such as by performing a multi-factor comprehensive evaluation of the simulation operation state based on fuzzy logic to calculate the safety coefficient of the simulation operation state.

[0072] This invention, through the generation of decision commands for the energy storage device, guides the device's charging and discharging behavior, power adjustment, and mode switching during actual operation, achieving an optimal balance between economy, stability, and environmental adaptability while ensuring safety. Specifically, the decision commands refer to the operational commands generated by the decision-making layer within the energy storage device's intelligent control system, based on a comprehensive analysis of multi-dimensional information such as the probability of extreme environmental events, equipment operational risks, simulation results, and feasibility assessments, to guide the actual operation of the energy storage device.

[0073] S5. Based on the decision instruction, the operating parameters of the energy storage device are adjusted through the execution layer to perform regulation of the energy storage device under the extreme environment.

[0074] This invention, through the execution layer adjusting the operating parameters of the energy storage device based on the decision-making instructions, can transform abstract decision-making instructions into specific device operations, enabling the energy storage device to operate precisely according to the requirements of the decision-making instructions. The operating parameters refer to various control variables and state variables, such as charging / discharging power, charging / discharging duration, and start / stop time, that are adjusted or set in real time by the execution layer according to the decision-making instructions during the operation of the energy storage device.

[0075] The embodiments of the present invention can ensure that the energy storage device can still provide stable service under extreme conditions by performing regulation of the energy storage device in the extreme environment, thereby improving the overall energy storage device's ability to withstand extreme events.

[0076] Optionally, to verify the effectiveness of the method of the present invention, simulation tests were conducted on an energy storage experimental platform. The method of the present invention was used for regulation under simulated extreme high temperature (45℃) and high humidity (90%RH) environments. The accuracy of the control commands reached 98%, which is about 25% higher than the traditional threshold control method; The system failure rate was reduced by 30%, and equipment operating efficiency was increased by 15%. Under the federated learning framework, the model convergence speed is improved by 20%, and the data privacy of each node is effectively protected.

[0077] This method has been piloted in actual energy storage power stations. The measured data shows that the system can still maintain stable operation during the passage of typhoons, with a control response time of less than 500ms, which significantly improves the resilience and adaptability of the energy storage system under extreme environments.

[0078] Compared to the problems described in the background technology, this invention constructs a global energy storage regulation model using federated learning technology. This model effectively integrates data resources from multiple sources, improving the model's generalization ability and privacy protection. The model comprises a perception layer, an analysis layer, a decision-making layer, and an execution layer, forming a complete closed-loop intelligent regulation system. The perception layer collects real-time meteorological data, equipment operating status data, and historical fault database data to construct a multimodal time-series dataset, forming an environmental perception vector that comprehensively characterizes the external environment and internal state of the energy storage equipment. Based on this vector, the analysis layer accurately predicts the probability of extreme environmental events and equipment operating risks, providing a scientific basis for decision-making. The decision-making layer generates preliminary regulation commands, evaluates their feasibility and safety through simulation, and ultimately outputs reliable decision commands, ensuring the scientific rationality of the regulation strategy. The execution layer dynamically adjusts equipment operating parameters according to the decision commands, achieving precise regulation of energy storage equipment in extreme environments. This invention not only improves the adaptability and safety of energy storage systems in complex environments but also significantly optimizes energy dispatch efficiency and equipment lifespan, providing key technical support for building an intelligent and resilient energy internet and promoting the intelligent and sustainable development of energy systems. Therefore, the extreme environment energy storage regulation method based on the interconnected large model provided in this embodiment of the invention can improve the accuracy and adaptability of extreme environment energy storage regulation.

[0079] like Figure 3 The diagram shown is a functional block diagram of an extreme environment energy storage and control system based on an interconnected large model according to the present invention.

[0080] The extreme environment energy storage and regulation system 300 based on an interconnected large-scale model described in this invention can be installed in an electronic device. Depending on the functions implemented, the extreme environment energy storage and regulation system based on an interconnected large-scale model can regulate the large-scale model construction module 301, the environmental sensing module 302, the risk analysis module 303, the instruction generation module 304, and the energy storage and regulation module 305. The module described in this invention can also be called a unit, referring to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0081] In this embodiment of the invention, the functions of each module / unit are as follows: The large-scale regulation model construction module 301 is used to acquire the extreme environmental characteristics of the energy storage device to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, a global energy storage regulation model of the energy storage device is constructed. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer. The environmental perception module 302 is used to collect multimodal time-series datasets of energy storage devices under extreme environments through the perception layer. The multimodal time-series datasets include real-time meteorological data, equipment operating status data, and historical fault database data to construct the environmental perception vector of the energy storage device. The risk analysis module 303 is used to analyze the probability of extreme environmental events and the operational risks of the energy storage device through the analysis layer based on the environmental perception vector. The instruction generation module 304 is used to generate control instructions for the energy storage device through the decision layer based on the probability of the extreme environmental event and the operational risk of the device, simulate the control instructions to obtain simulation results, and analyze the executability and security of the control instructions based on the simulation results, so as to finally generate decision instructions for the energy storage device. The energy storage control module 305 is used to adjust the operating parameters of the energy storage device through the execution layer based on the decision command, so as to perform the control of the energy storage device in the extreme environment.

[0082] In detail, the modules in the extreme environment energy storage and control system 200 based on an interconnected large model described in this embodiment of the invention adopt the same characteristics as described above during use. Figure 1 The method is the same as the extreme environment energy storage regulation method based on the interconnected large model described in the article, and can produce the same technical effect, so it will not be repeated here.

[0083] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0084] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for extreme environment energy storage regulation based on an interconnected large-scale model, characterized in that, The method includes: The extreme environmental characteristics of the energy storage device are obtained to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, a global energy storage regulation model of the energy storage device is constructed. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer. The sensing layer collects multimodal time-series datasets of energy storage devices under extreme environments, including real-time meteorological data, equipment operating status data, and historical fault database data, to construct the environmental perception vector of the energy storage device. Based on the environmental perception vector, the analysis layer analyzes the probability of extreme environmental events and the operational risks of the energy storage device. Based on the probability of extreme environmental events and the operational risks of the equipment, the decision-making layer initially generates control commands for the energy storage device, simulates the control commands, obtains simulation results, and analyzes the executability and security of the control commands based on the simulation results, so as to finally generate decision commands for the energy storage device. Based on the decision instructions, the operating parameters of the energy storage device are adjusted through the execution layer to perform regulation of the energy storage device under the extreme environment.

2. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, The construction of a global energy storage regulation model for the energy storage device based on the federated learning environment includes: Configure the sensors of the energy storage device to construct a sensor network for the energy storage device; Based on the sensor network, the sensing layer of the energy storage device is determined; Based on the federated learning environment, a global environment analysis federated model of the energy storage device is constructed to determine the analysis layer of the energy storage device. Define the control objectives of the energy storage device in order to construct the target excitation function of the energy storage device; Based on the target incentive function, the decision layer of the energy storage device is constructed; Determine the PLC algorithm of the energy storage device to construct the execution layer of the energy storage device; By integrating the perception layer, the analysis layer, the decision-making layer, and the execution layer, a global energy storage regulation model is obtained.

3. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 2, characterized in that, The construction of a global environment analysis federated model for the energy storage device based on the federated learning environment includes: Obtain the local dataset corresponding to the local node in the federated learning environment; The global environment analysis model corresponding to the federated learning environment is deployed to the local node so as to train the global environment analysis model using the local dataset to obtain the local environment analysis model. Extract the model parameters of the local environment analysis model; The model parameters are aggregated to obtain aggregated parameters; The global environment analysis model is updated using the aggregation parameters to obtain a global environment analysis federated model.

4. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 2, characterized in that, The target excitation function for constructing the energy storage device includes: The target indicators corresponding to the regulation target of the energy storage device are quantified, wherein the target indicators include: safety indicators, economic indicators, stability indicators and lifespan indicators; Based on the target indicators, a plurality of incentive functions are constructed for the energy storage device, wherein the plurality of incentive functions include: a safety incentive function, an economic incentive function, a stability incentive function, and a lifetime incentive function; Determine the regulation sensitivity of the energy storage device in order to calculate the dynamic multinomial weights of the multinomial excitation functions; Calculate the extreme environmental factors of the energy storage device under the extreme environment; The target activation function is obtained by weighting the multinomial activation functions based on the dynamic multinomial weights and the extreme environmental factors.

5. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, The federated learning environment simulating energy storage devices under extreme environments includes: Determine the federated learning framework for the energy storage device; Determine the central server of the federated learning framework and the local nodes of the energy storage device; Establish the communication protocol and encryption algorithm between the central server and the local nodes; Based on the extreme environmental characteristics corresponding to the energy storage device, a global environment analysis model of the federated learning framework is defined. The federated learning environment of the energy storage device is determined based on the communication protocol, the encryption algorithm, and the global environment analysis model.

6. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, The construction of the environmental perception vector for the energy storage device includes: The multimodal time series dataset corresponding to the energy storage device is preprocessed to obtain a preprocessed multimodal time series dataset; Extract multimodal data features from the preprocessed multimodal time-series dataset; The multimodal data features are fused to obtain high-dimensional fused features; The high-dimensional fusion features are reduced in dimensionality to obtain the dimensionality-reduced fusion features; Based on the aforementioned dimensionality reduction and fusion features, an environmental perception vector for the energy storage device is constructed.

7. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, The process of analyzing the probability of extreme environmental events and the operational risks of the energy storage device based on the environmental perception vector through the analysis layer includes: Extract the meteorological feature sequence and equipment operating status sequence of the environmental perception vector; Calculate the probability of extreme environmental events for the energy storage device based on the meteorological characteristic sequence; Based on the probability of the extreme environmental events, identify the operational risk indicators of the energy storage device; The operational risk of the energy storage device is calculated based on the operational status sequence and the operational risk indicators.

8. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, Based on the probability of extreme environmental events and the operational risks of the equipment, the decision-making layer initially generates control instructions for the energy storage equipment, including: Calculate the comprehensive risk index of the energy storage device based on the probability of the extreme environmental events and the operational risks of the device. Based on the comprehensive risk index, the potential risk factors of the energy storage device are determined; Based on the potential risk factors, the instruction generation logic of the energy storage device is determined; Based on the instruction generation logic, the control instructions for the energy storage device are initially generated.

9. The extreme environment energy storage regulation method based on the interconnected large model as described in claim 1, characterized in that, The analysis of the executability and security of the control commands includes: The simulation results corresponding to the control commands are normalized to obtain normalized simulation results; Extract the equipment state characteristics, system response characteristics, and environmental parameters from the normalized simulation results; Based on the device status characteristics, determine the simulated operating status of the energy storage device corresponding to the control command; Calculate the safety coefficient of the simulation running state to determine the safety of the control commands; Based on the system response characteristics, analyze the timing feasibility of the control command; Based on the environmental parameters, the environmental adaptability of the control command is determined; The executability of the control command is determined by considering the aforementioned security, time feasibility, and environmental adaptability.

10. An extreme environment energy storage and control system based on an interconnected large-scale model, characterized in that, The system includes: The large-scale regulation model construction module is used to acquire the extreme environmental characteristics of the energy storage device to simulate the federated learning environment of the energy storage device under extreme conditions. Based on the federated learning environment, a global energy storage regulation model of the energy storage device is constructed. The global energy storage regulation model includes: a perception layer, an analysis layer, a decision layer, and an execution layer. An environmental perception module is used to collect multimodal time-series datasets of energy storage devices under extreme environments through the perception layer. The multimodal time-series datasets include real-time meteorological data, equipment operating status data, and historical fault database data to construct the environmental perception vector of the energy storage device. The risk analysis module is used to analyze the probability of extreme environmental events and the operational risks of the energy storage device through the analysis layer based on the environmental perception vector. The instruction generation module is used to generate preliminary control instructions for the energy storage device through the decision layer based on the probability of the extreme environmental events and the operational risks of the device, simulate the control instructions to obtain simulation results, and analyze the executability and security of the control instructions based on the simulation results, so as to finally generate decision instructions for the energy storage device. The energy storage regulation module is used to adjust the operating parameters of the energy storage device through the execution layer based on the decision command, so as to perform regulation of the energy storage device in the extreme environment.

Images