All-vanadium redox flow energy storage battery management system and method based on situation awareness

Abstract

The invention discloses an all-vanadium redox flow energy storage battery management system and method based on situation awareness, and relates to the technical field of energy storage battery management. An all-vanadium redox flow energy storage battery management system multi-physics field digital twinborn model is constructed, core parameters are configured, and a situation awareness module, a sensor and an actuator are initialized; an electrolyte concentration method, a coulomb counting method and extended Kalman filtering are fused to estimate SOC, SOH, a polarization state and a thermal management state, and a fuzzy comprehensive evaluation method is adopted to identify an operation situation. According to the method, the multi-physics field digital twin model is coupled with electrochemical, heat conduction and fluid dynamic characteristics, and a state estimation strategy of electrolyte concentration method-coulomb counting method fusion + extended Kalman filtering is combined, so that the estimation accuracy of key state quantities such as SOC and SOH is remarkably improved, and a reliable foundation is laid for situation awareness.

Description

A situational awareness-based all-vanadium redox flow battery management system and method Technical Field

[0001] This invention relates to the field of energy storage battery management technology, and in particular to a situational awareness-based all-vanadium redox flow battery management system and method. Background Technology

[0002] With the rapid development of the new energy industry, energy storage technology is becoming increasingly important as a key support for mitigating fluctuations in new energy output and enhancing the grid's peak-shaving capacity. Vanadium redox flow batteries (VRBs) have gained widespread attention and application in large-scale energy storage due to their significant advantages, including independently adjustable capacity, long cycle life, high safety, and environmental friendliness. However, VRB systems are complex systems involving the coupling of multiple physical fields such as electrochemistry, heat conduction, and fluid dynamics. Their operating state is dynamically affected by multiple factors, including electrolyte concentration, flow rate, temperature, charge / discharge power, and the external grid environment. Traditional battery management methods are no longer sufficient to meet the system's requirements for high precision, high reliability, and economic efficiency.

[0003] A vanadium redox flow battery management system (BMS) is a comprehensive management system for energy storage battery systems. Its main functions include controlling the redox flow circulation system, nitrogen charging protection system, battery charging and discharging, air conditioning and cooling system, and electrical protection system. Traditional BMS systems are mainly management systems that realize charging and discharging functions, and do not have functions such as early fault prediction, equipment operation trend analysis, and early warning of equipment failure. They are more about remedial handling after a fault occurs to reduce the scope of the accident.

[0004] Therefore, this invention provides a situational awareness-based vanadium redox flow battery management system. The situational awareness system is a comprehensive system that uses widely deployed sensors, information interaction, and intelligent analysis platforms to monitor the real-time operating status of equipment, assess its health level, and predict future risks, ultimately providing decision support for the safe, stable, and efficient operation of the equipment. This situational awareness-based vanadium redox flow battery management system enables energy storage battery systems to evolve from a "passive response" mode relying on human experience to a "proactive early warning and intelligent decision-making" mode based on data and algorithms. This is a key step in realizing the digital and intelligent transformation of energy storage battery systems. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a situational awareness-based vanadium redox flow battery management system and method. The technical solution adopted is as follows: A situational awareness-based vanadium redox flow battery management method, comprising the following steps: Step 1, constructing a multiphysics digital twin model of the vanadium redox flow battery management system, configuring core parameters, and initializing the situational awareness module, sensors, and actuators; Step 2, deploying a sensor network to collect internal battery state data and integrating API interfaces to acquire external environment data; Step 3, processing data using adaptive wavelet transform and an improved Z-score algorithm, achieving multi-source data fusion through federated Kalman filtering, and extracting time-series features; Step 4, fusing electrolyte concentration method, coulomb counting method, and extended Kalman filter to estimate... Step 5: Based on the SOC, SOH, polarization state, and thermal management state, the fuzzy comprehensive evaluation method is used to identify the operating status; Step 6: Based on the attention mechanism LSTM, the short-term grid and battery state is predicted, and the long-term performance and electricity price are predicted by combining the improved Arrhenius model and the ARIMA-GARCH model; Step 7: A multi-objective optimization function is constructed, and the MPC algorithm is used to dynamically optimize the electrolyte flow rate, charge and discharge power, thermal management, and economic dispatch strategy; Step 8: The optimization instructions are verified for safety, converted into control signals for execution, and the execution deviation is monitored; Step 9: The RLS algorithm is used to update the model parameters online to achieve real-time synchronization of the digital twin model.

[0006] Optionally, the multiphysics digital twin model in step 1 includes an electrochemical model, a heat conduction model, and a fluid dynamics model, with core parameters including electrolyte concentration, rated power, temperature safety threshold, SOC operating window, and economic parameters.

[0007] Optionally, the battery internal status data in step 2 includes electrolyte flow rate, concentration, temperature, battery stack voltage, current, internal resistance, and equipment operating status; the external environment data includes grid load, renewable energy output, time-of-use electricity price, and ambient temperature and humidity.

[0008] Optionally, in step 4, the electrolyte concentration method and the coulomb counting method are used to make a preliminary estimate of the SOC, and the results of the two methods are fused through an adaptive weighting factor. The core fusion formula is as follows: ;in An adaptive weighting factor that is dynamically adjusted based on sensor reliability. SOC calculated using the electrolyte concentration method. The SOC is calculated based on the Coulomb integral; the fusion results will be used. As the observation input, the extended Kalman filter achieves the optimal joint estimation of charge state, health state, polarization state and thermal management state; the fuzzy comprehensive evaluation method is adopted, and the optimal joint estimation is used as the evaluation factor. Through fuzzy synthesis operation and the maximum membership principle, the comprehensive operating status level of the system is identified.

[0009] Optionally, for short-term forecasting, a long short-term memory network model with an attention mechanism is used to predict short-term grid demand and battery state trends. The attention mechanism focuses the model on key time-series information to predict short-term grid demand and battery state trends. For long-term forecasting, an improved Arrhenius model is used to predict battery life degradation, and an ARIMA-GARCH hybrid model is used to predict electricity prices and their fluctuations.

[0010] Optionally, step 6 constructs a multi-objective optimization function for model predictive control rolling, with the core form as follows: ;in , and As a weighting factor, , and These are sub-objective functions for efficiency, lifetime decay, and economic cost, respectively. In each control cycle, based on the function and the prediction model, the optimal control sequence that satisfies all safety constraints in the future time domain is solved, and the first term of the sequence is executed to achieve dynamic optimization of electrolyte flow rate, charging and discharging power, and thermal management.

[0011] Optionally, step 7 performs a security check on the optimization instructions; if the instructions... If a constraint is violated, the projective method is used to constrain it to the nearest safe boundary, as shown below: ;in For optimized quality, and To control the upper and lower limits of the safety of the variables, the safety instructions are converted into specific actuator signals through calibration relationships, and the execution deviation is monitored. When the threshold is exceeded, feedback adjustment is triggered.

[0012] Optionally, step 8 specifically includes: updating the parameters of the digital twin model online using a recursive least squares method with a forgetting factor λ. The core parameter update formula is: ;in, Here is the gain matrix. These are actual observed values. The observation vector is used to update the model parameters in real time to track the time-varying characteristics of the battery system, keeping the digital twin model synchronized with the physical entity.

[0013] Optionally, step 9 is also included: based on the situation assessment and prediction results, automatically generate a system health status report and maintenance recommendations, and perform real-time data visualization, situation level display and multi-level alarm information push through the human-computer interaction interface; when key performance indicators are lower than preset thresholds, an early warning is automatically triggered in the maintenance recommendations.

[0014] A situational awareness-based vanadium redox flow battery management system is disclosed to implement a situational awareness-based management method for vanadium redox flow batteries. The system comprises a digital twin model management module, a multi-source data acquisition and fusion module, a situational awareness and prediction module, a predictive analysis unit, an intelligent decision-making and optimization module, a safety control and execution module, a model online update module, and a human-computer interaction and reporting module. The digital twin model management module is used to construct and maintain a multi-physics digital twin model of the vanadium redox flow battery, and to perform model initialization and parameter configuration. The multi-source data acquisition and fusion module is used to acquire real-time internal state data and external environmental data of the battery, and to perform data preprocessing and fusion. The situational awareness and prediction module is used to fuse electrolyte concentration method, coulomb counting method, and extended Kalman filter to estimate SOC, SOH, polarization state, and thermal properties in real time. The system is managed in several ways: a fuzzy comprehensive evaluation method is used to assess the system's operational status; a predictive analysis unit uses an attention-based LSTM model for short-term state prediction and combines an improved Arrhenius model with an ARIMA-GARCH model for long-term performance and market prediction; an intelligent decision-making and optimization module constructs multi-objective optimization functions and uses model predictive control algorithms to dynamically generate optimization strategies for electrolyte flow, charging and discharging power, thermal management, and economic scheduling; a safety control and execution module performs safety verification on optimization commands and converts them into control signals to drive actuators, while monitoring execution deviations; a model online update module uses a recursive least squares algorithm to update the digital twin model parameters in real time, ensuring synchronization between the model and the physical entity; and a human-computer interaction and reporting module visualizes real-time status data, generates health reports, and issues maintenance suggestions and management alarms.

[0015] In summary, the present invention includes at least one of the following beneficial technical effects: The present invention can provide a situational awareness-based all-vanadium redox flow battery management system and method. By coupling electrochemical, thermal conduction and fluid dynamics characteristics through a multi-physics digital twin model, and combining the state estimation strategy of electrolyte concentration method-Coulomb counting method fusion + extended Kalman filter, the estimation accuracy of key state quantities such as SOC and SOH is significantly improved, laying a reliable foundation for situational awareness.

[0016] By employing an attention mechanism LSTM to focus on key short-term time-series information, and combining an improved Arrhenius model with an ARIMA-GARCH model, long-term performance and electricity price co-prediction are achieved, providing comprehensive predictive input for multi-objective optimization and enhancing the foresight of decision-making.

[0017] A multi-objective optimization function integrating efficiency, lifespan, and cost is constructed. Through model predictive control, electrolyte flow rate, charge / discharge power, and thermal management strategies are dynamically optimized to balance system operation economy, safety, and durability, achieving global optimum.

[0018] The digital twin model parameters are updated online using the recursive least squares method with a forgetting factor, and the time-varying characteristics of the battery are tracked in real time to ensure that the model is always highly synchronized with the physical entity and maintain long-term stable management results. Attached Figure Description

[0019] Figure 1 is a flowchart illustrating a situational awareness-based management method for all-vanadium redox flow batteries according to the present invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings.

[0021] This invention discloses a situational awareness-based all-vanadium redox flow battery management system and method.

[0022] Referring to Figure 1, Example 1, a situational awareness-based management method for a vanadium redox flow battery includes the following steps: Step 1, constructing a multi-physics digital twin model of the vanadium redox flow battery management system, configuring core parameters, and initializing the situational awareness module, sensors, and actuators; Step 2, deploying a sensor network to collect internal battery state data and integrating API interfaces to acquire external environment data; Step 3, processing data using adaptive wavelet transform and an improved Z-score algorithm, achieving multi-source data fusion through federated Kalman filtering, and extracting time-series features; Step 4, fusing electrolyte concentration method, coulomb counting method, and extended Kalman filter to estimate... Step 5: Based on the SOC, SOH, polarization state, and thermal management state, the fuzzy comprehensive evaluation method is used to identify the operating status; Step 6: Based on the attention mechanism LSTM, the short-term grid and battery state is predicted, and the long-term performance and electricity price are predicted by combining the improved Arrhenius model and the ARIMA-GARCH model; Step 7: A multi-objective optimization function is constructed, and the MPC algorithm is used to dynamically optimize the electrolyte flow rate, charge and discharge power, thermal management, and economic dispatch strategy; Step 8: The optimization instructions are verified for safety, converted into control signals for execution, and the execution deviation is monitored; Step 9: The RLS algorithm is used to update the model parameters online to achieve real-time synchronization of the digital twin model.

[0023] Example 2: In step 1, the multiphysics digital twin model includes an electrochemical model, a heat conduction model, and a fluid dynamics model. The core parameters include electrolyte concentration, rated power, temperature safety threshold, SOC operating window, and economic parameters.

[0024] Example 3: In step 2, the internal state data of the battery includes electrolyte flow rate, concentration, temperature, battery stack voltage, current, internal resistance, and equipment operating status; the external environmental data includes grid load, new energy output, time-of-use electricity price, and ambient temperature and humidity.

[0025] By adopting the above technical solution, a multi-physics digital twin model is constructed to integrate the multi-physics characteristics involved in battery operation. At the same time, core parameters are configured and situational awareness modules, sensors and actuators are initialized, providing a basic model framework and hardware-software collaboration foundation for the entire management method.

[0026] Deploying a sensor network can directly collect various status data during the battery's internal operation, while integrating API interfaces can access relevant data from the external environment, enabling comprehensive acquisition of both internal and external data.

[0027] Adaptive wavelet transform can effectively suppress noise and interference in the data, the improved Z-score algorithm can identify and process abnormal data, and federated Kalman filtering can fuse data from different sources, reduce data uncertainty, and extract time-series features that reflect the operating rules of the system.

[0028] The fusion of electrolyte concentration method and coulomb counting method can leverage the advantages of both methods to achieve preliminary SOC estimation. Combined with extended Kalman filtering, it can accurately estimate SOC, SOH, polarization state and thermal management state. The fuzzy comprehensive evaluation method can effectively identify the system operation status level through comprehensive analysis of various state variables.

[0029] The attention-based LSTM model enables the model to focus on short-term time-series information that significantly affects the prediction results, thereby improving the accuracy of short-term power grid and battery state predictions. The improved Arrhenius model can characterize the decay of battery life over time and with the environment, while the ARIMA-GARCH model can take into account both the trend and volatility of electricity prices, achieving reliable predictions of long-term performance and electricity prices.

[0030] Constructing a multi-objective optimization function can take into account various requirements such as efficiency, lifespan, and economy. The MPC algorithm achieves dynamic optimization of electrolyte flow rate, charging and discharging power, etc. by solving for the optimal control sequence in the future time domain and executing the first term in each control cycle.

[0031] Performing safety checks on optimized instructions can prevent them from exceeding safe limits. Converting them into control signals for execution and monitoring deviations can ensure that control actions are accurately implemented and correct execution deviations in a timely manner.

[0032] The RLS algorithm can be used to update the parameters of the digital twin model online using the latest observation data, enabling the model to track the time-varying characteristics of the battery system in real time and maintain synchronization with the physical entity.

[0033] The multiphysics digital twin model includes an electrochemical model that describes the electrochemical reaction process inside the battery, a thermal conduction model that reflects the heat transfer law of the battery, and a fluid dynamics model that characterizes the electrolyte flow characteristics. The combination of these three models enables a comprehensive simulation of the multiphysics coupling process of the battery. Among the core parameters, electrolyte concentration affects electrochemical reaction efficiency, rated power determines the upper limit of battery operating power, temperature safety threshold ensures that the battery avoids overheating damage, SOC operating window defines the safe range of battery charging and discharging, and economic parameters provide a basis for the formulation of economic scheduling strategies. Together, they support the accurate operation and management decisions of the model.

[0034] Internal battery status data, including electrolyte flow rate, concentration, and temperature, directly reflects the electrolyte's operating status. Battery stack voltage, current, and internal resistance reflect the battery's electrical performance. Equipment operating status reflects the system hardware's operational status. These data collectively constitute a complete picture of the battery's own operating state. External environmental data, such as grid load and renewable energy output, influences battery charging and discharging scheduling demands. Time-of-use pricing relates to economic dispatch strategies, and ambient temperature and humidity affect battery performance. The synergy of internal and external data provides comprehensive and effective data support for subsequent data processing, status estimation, and optimization decisions.

[0035] Example 4: In step 4, the electrolyte concentration method and coulomb counting method are used to make a preliminary estimate of SOC, and the results of the two are fused through an adaptive weighting factor. The core fusion formula is: ;in An adaptive weighting factor that is dynamically adjusted based on sensor reliability. SOC calculated using the electrolyte concentration method. The SOC is calculated based on the Coulomb integral; the fusion results will be used. As the observation input, the extended Kalman filter achieves the optimal joint estimation of charge state, health state, polarization state and thermal management state; the fuzzy comprehensive evaluation method is adopted, and the optimal joint estimation is used as the evaluation factor. Through fuzzy synthesis operation and the maximum membership principle, the comprehensive operating status level of the system is identified.

[0036] By adopting the above technical solutions, the electrolyte concentration method calculates SOC based on the intrinsic correlation between vanadium ion concentration in the electrolyte and SOC, while the coulomb counting method obtains SOC by integrating the charge and discharge current. The fusion of these two methods leverages their respective advantages under different operating conditions to achieve a preliminary SOC estimate. An adaptive weighting factor is dynamically adjusted based on the reliability of sensor-acquired data, making the fusion result more closely reflect actual operating conditions. Using this fusion result as the observation input, the extended Kalman filter utilizes its filtering characteristics to handle uncertainties in state estimation, achieving optimal joint estimation of charge state, health state, polarization state, and thermal management state. The fuzzy comprehensive evaluation method uses the state variables obtained from the optimal joint estimation as evaluation factors. Through fuzzy synthesis operations, the influence of each factor on the situation is integrated, and then the overall system operating status level is determined based on the maximum membership principle, achieving accurate judgment of the operating status.

[0037] Example 5: For short-term forecasting, a long short-term memory network model with an attention mechanism is used to predict short-term grid demand and battery status trends. The attention mechanism focuses the model on key time-series information to predict short-term grid demand and battery status trends. For long-term forecasting, an improved Arrhenius model is used to predict battery life degradation, and an ARIMA-GARCH hybrid model is used to predict electricity prices and their fluctuations.

[0038] By adopting the above technical solutions, for short-term forecasting, the Long Short-Term Memory (LSTM) network model itself has the ability to capture long-term dependencies in time-series data. The attention mechanism calculates the importance of different time-series information to the prediction target, enabling the model to prioritize key time-series information that significantly affects short-term grid demand and battery state trends, reducing interference from irrelevant information and thus improving the accuracy of short-term forecasts. For long-term forecasting, the improved Arrhenius model, by incorporating actual operating parameters during battery operation, can more accurately characterize the impact of factors such as temperature and depth of charge / discharge on battery life degradation, achieving reliable prediction of long-term battery life degradation trends. In the ARIMA-GARCH hybrid model, the ARIMA model can capture the linear trend and periodic changes in electricity prices, while the GARCH model can effectively characterize the clustering and heteroscedasticity of electricity price fluctuations. The combination of the two can comprehensively reflect the long-term change characteristics and fluctuation features of electricity prices, achieving accurate prediction of electricity prices and their fluctuations.

[0039] Example 6, Step 6: Construct a multi-objective optimization function for model predictive control rolling, the core form of which is: ;in , and As a weighting factor, , and These are sub-objective functions for efficiency, lifetime decay, and economic cost, respectively. In each control cycle, based on the function and the prediction model, the optimal control sequence that satisfies all safety constraints in the future time domain is solved, and the first term of the sequence is executed to achieve dynamic optimization of electrolyte flow rate, charging and discharging power, and thermal management.

[0040] By adopting the above technical solution, the weighting factors in the constructed multi-objective optimization function , and These sub-objective functions respectively reflect the importance of efficiency, lifespan decay, and economic cost in the optimization objective. Adjusting the weights allows for different objective emphases under varying management needs. , and By quantifying system operating efficiency, battery life degradation, and economic costs separately, the integrated optimization goal formed by combining these three factors can take into account multi-dimensional needs and avoid system imbalance caused by optimizing a single goal.

[0041] Model predictive control predicts the system state for a future period within each control cycle by combining the predictive model. It then solves for the optimal control sequence that satisfies the safety constraints within the time domain based on a multi-objective optimization function. After executing only the first term of the sequence, it enters the next control cycle to re-predict and solve. Through this rolling optimization method, it can respond to changes in system state in real time and achieve dynamic optimization and adjustment of electrolyte flow rate, charging and discharging power, and thermal management.

[0042] Example 7: In step 7, a security check is performed on the optimization instruction. If the instruction... If a constraint is violated, the projective method is used to constrain it to the nearest safe boundary, as shown below: ;in For optimized quality, and To control the upper and lower limits of the safety of the variables, the safety instructions are converted into specific actuator signals through calibration relationships, and the execution deviation is monitored. When the threshold is exceeded, feedback adjustment is triggered.

[0043] By adopting the above technical solution, the security verification of the optimization command in step 7 is to prevent the command from exceeding the system's safe operating range and to ensure the stability of the device and system. When the command... When constraints are violated, the projection method uses a first-take approach. and The maximum value is determined to be no less than the lower limit, and then that value is compared with... The minimum value will constrain the instruction to. and Within the established safety boundary, safety instructions are received. .

[0044] By calibrating the relationship, safety commands are converted into specific signals that the actuators can recognize, ensuring that control commands can accurately drive the actuators to move. At the same time, deviations during the execution process are monitored in real time. When the deviation exceeds a set threshold, a feedback adjustment mechanism is triggered to correct the execution deviation in a timely manner and maintain the system's stable operation according to the optimization goal.

[0045] Example 8, step 8 specifically includes: updating the parameters of the digital twin model online using a recursive least squares method with a forgetting factor λ. The core parameter update formula is: ;in, Here is the gain matrix. These are actual observed values. The observation vector is used to update the model parameters in real time to track the time-varying characteristics of the battery system, keeping the digital twin model synchronized with the physical entity.

[0046] By adopting the above technical solution, the recursive least squares method with a forgetting factor λ assigns higher weight to recent observation data, reducing the impact of earlier data on parameter updates, thus enabling the algorithm to better adapt to the time-varying characteristics of the battery system. In the core parameter update formula, The Kk gain matrix represents the model parameters from the previous time step and is used to adjust the parameter update magnitude. and The difference reflects the deviation between the actual observed value and the model prediction value. The parameters at the previous time step are corrected by multiplying this deviation by the gain matrix to obtain the model parameters at the current time step. Through this online update method, the model parameters can track the time-varying characteristics of the battery system caused by electrolyte aging, component wear, etc., in real time, ensuring that the digital twin model and the physical entity always remain synchronized and maintain the model's ability to accurately depict the system's operating state.

[0047] Example 9 also includes step 9: Based on the situation assessment and prediction results, automatically generate a system health status report and maintenance suggestions, and perform real-time data visualization, situation level display and multi-level alarm information push through the human-computer interaction interface; when the key performance indicators are lower than the preset threshold, automatically trigger an early warning in the maintenance suggestions.

[0048] Example 10: A situational awareness-based vanadium redox flow battery management system, used to implement a situational awareness-based vanadium redox flow battery management method. The system is characterized by comprising a digital twin model management module, a multi-source data acquisition and fusion module, a situational awareness and prediction module, a predictive analysis unit, an intelligent decision-making and optimization module, a safety control and execution module, a model online update module, and a human-computer interaction and reporting module. The digital twin model management module is used to construct and maintain a multi-physics digital twin model of the vanadium redox flow battery, and to perform model initialization and parameter configuration. The multi-source data acquisition and fusion module is used to acquire real-time battery internal state data and external environmental data, and to perform data preprocessing and fusion. The situational awareness and prediction module is used to fuse electrolyte concentration method, coulomb counting method, and extended Kalman filter to estimate SOC, SOH, and polarization state in real time. The system is divided into several modules: a dynamic and thermal management status module; a fuzzy comprehensive evaluation method is used to assess the system's operational status; a predictive analysis unit uses an attention-based LSTM model for short-term state prediction and combines an improved Arrhenius model and an ARIMA-GARCH model for long-term performance and market prediction; an intelligent decision-making and optimization module constructs a multi-objective optimization function and uses a model predictive control algorithm to dynamically generate optimization strategies for electrolyte flow, charging and discharging power, thermal management, and economic scheduling; a safety control and execution module performs safety verification on optimization commands and converts them into control signals to drive actuators, while monitoring execution deviations; a model online update module uses a recursive least squares algorithm to update the digital twin model parameters in real time to ensure synchronization between the model and the physical entity; and a human-computer interaction and reporting module visualizes real-time status data, generates health reports, and issues maintenance suggestions and management alarm information.

[0049] The following specific embodiments illustrate the implementation principle of the present invention: Taking a 10MW / 40MWh all-vanadium redox flow storage power station as the application object, the specific implementation is as follows: Step 1: Model construction and initialization. Construct a multi-physics digital twin model. The electrochemical model adopts the equivalent circuit model corrected by the Nernst equation. The thermal conduction model sets the thermal conductivity of the battery stack to 0.8W / (m·K) and the specific heat capacity of the electrolyte to 3800J / (kg·K). The fluid dynamics model describes the laminar flow state of the electrolyte in the bipolar plate channel. The core parameters are configured as follows: initial electrolyte concentration 1.8mol / L, rated power 10MW, temperature safety threshold 50℃, SOC working window 20%-80%, peak-valley electricity price difference 0.5 yuan / (kWh). Initialize the sampling frequency of the situational awareness module to 10Hz, and deploy voltage sensors, current sensors, temperature sensors and flow sensors. The actuators include electrolyte pumps and charge-discharge controllers.

[0050] Step 2: Data Acquisition. Sensor network collects internal battery status data: electrolyte flow rate, concentration, battery stack inlet and outlet temperatures, voltage, current, internal resistance, and pump operating status; external environmental data is obtained through the power grid dispatch center API interface: power grid load, photovoltaic and wind power output, time-of-use electricity price, and ambient temperature and humidity.

[0051] Step 3: Data processing and fusion. Adaptive wavelet transform based on db4 wavelet basis is used to decompose and denoise voltage and current data in three layers. The improved Z-score algorithm is set with a threshold of 3.0 to identify abnormal temperature data. The federated Kalman filter is set with 5 fusion nodes to process voltage, current, temperature, flow rate and external data respectively. After fusion, the time series features such as battery charging and discharging power fluctuation period and temperature change trend are extracted.

[0052] Step 4: State Estimation and Situation Identification. Electrolyte Concentration Method: The State of Charge (SOC) is calculated using data from a vanadium ion concentration detector. The SOC is obtained by integrating the charge and discharge current using the coulomb counting method. The adaptive weighting factor α is dynamically adjusted based on the error variance of the concentration sensor and the current sensor (range 0.3-0.7). The fused SOC is used as the observation input, and an extended Kalman filter is used to construct an 8-dimensional state vector (including SOC, State of Hypoxia (SOH), polarization voltage, and temperature of each region) to achieve joint estimation. Fuzzy Comprehensive Evaluation Method: The weights of SOC, SOH, temperature, and polarization state are set to 0.4, 0.2, 0.2, and 0.2, respectively. The state variables are fuzzified using a triangular membership function, and a weighted average method is used to synthesize a four-level operating situation rating: "Excellent, Good, Medium, and Poor".

[0053] Step 5: Situation Prediction. Short-term prediction uses an attention mechanism LSTM model with two hidden layers and a time step of 24 to predict the grid load and battery SOC trend for the next 4 hours. Long-term prediction uses an improved Arrhenius model with an activation energy of 80 kJ / mol to predict the battery SOH decay curve over 5 years. The ARIMA-GARCH model uses an ARIMA(2,1,1)-GARCH(1,1) structure to predict the time-of-use electricity price and fluctuation range for the next month.

[0054] Step 6: Optimize the scheduling multi-objective optimization function. Set weight factors w1=0.3 (efficiency), w2=0.4 (lifetime), w3=0.3 (cost), Je is the charge / discharge efficiency, Jl is the battery cycle life loss coefficient, and Jc is the time-of-use electricity price difference revenue. Set the prediction time domain to 12 (corresponding to 12 hours) and the control time domain to 3 (corresponding to 3 hours) to obtain the electrolyte flow rate (20-45m³). 3 The optimal control sequence for charge / h and charge / discharge power (-10MW to 10MW) is executed, with the first term of the sequence being executed.

[0055] Step 7: Safety Control and Safety Verification. Set the upper and lower limits of charge / discharge power to -10MW and 10MW, and the upper and lower limits of electrolyte flow rate to 15m³ / h. 3 / h, 50m 3 / h, the upper limit of temperature control is 45℃; the projection method is used to correct the command that exceeds the constraint to the safe boundary, and the command is sent to the actuator through a 4-20mA current signal to monitor the execution deviation. When the threshold is exceeded, the PID feedback adjustment is triggered.

[0056] Step 8: Model update. The electrochemical parameters (such as ohmic internal resistance and polarization resistance) of the digital twin model are updated online using the recursive least squares method with a forgetting factor λ=0.98. The gain matrix Kk is calculated every 10 minutes using the latest observation data, and the model parameters θk are corrected to ensure that the error between the model output and the physical entity is ≤5%.

[0057] Step 9: The operation and maintenance management generates a daily system health status report, including the SOH change curve, status level statistics, and remaining life prediction years; the human-machine interface displays the SOC, temperature, status level, and charge and discharge power curves in real time, and sets three levels of alarms: SOH≤80% triggers a warning, temperature≥48℃ triggers an alarm, and power deviation≥1MW triggers an emergency shutdown. Alarm information is pushed to operation and maintenance personnel via SMS and the platform.

[0058] System implementation configuration: The digital twin model management module is deployed on an industrial server and uses the Unity3D engine to build a visual model; the multi-source data acquisition and fusion module uses an edge computing gateway; the situational awareness and prediction module and the intelligent decision-making and optimization module run on a GPU server; the safety control and execution module uses a PLC controller; and the human-machine interaction and reporting module develops a web platform that supports access from PCs and mobile devices.

[0059] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A situational awareness-based management method for all-vanadium redox flow batteries, characterized in that, Includes the following steps: Step 1: Construct a multi-physics digital twin model of the vanadium redox flow battery management system, configure core parameters, and initialize the situational awareness module, sensors, and actuators; Step 2: Deploy a sensor network to collect internal battery state data and integrate API interfaces to obtain external environment data; Step 3: Process the data using adaptive wavelet transform and an improved Z-score algorithm, achieve multi-source data fusion through federated Kalman filtering, and extract time-series features. Step 4: Integrate electrolyte concentration method, coulomb counting method and extended Kalman filter to estimate SOC, SOH, polarization state and thermal management state, and use fuzzy comprehensive evaluation method to identify the operating status; Step 5: Based on attention mechanism LSTM, predict short-term grid and battery state, and combine improved Arrhenius model and ARIMA-GARCH model to predict long-term performance and electricity price; Step 6: Construct multi-objective optimization function, and use MPC algorithm to dynamically optimize electrolyte flow rate, charge and discharge power, thermal management and economic dispatch strategy; Step 7: Perform security verification on optimization instructions, convert them into control signals for execution and monitor execution deviation; Step 8: Use RLS algorithm to update model parameters online to achieve real-time synchronization of digital twin model.

2. The method for managing a vanadium redox flow battery based on situational awareness according to claim 1, characterized in that, The multiphysics digital twin model in step 1 includes an electrochemical model, a heat conduction model, and a fluid dynamics model. The core parameters include electrolyte concentration, rated power, temperature safety threshold, SOC operating window, and economic parameters.

3. The method for managing a vanadium redox flow battery based on situational awareness according to claim 2, characterized in that, In step 2, the internal battery status data includes electrolyte flow rate, concentration, temperature, battery stack voltage, current, internal resistance, and equipment operating status; the external environmental data includes grid load, renewable energy output, time-of-use electricity price, and ambient temperature and humidity.

4. The method for managing a vanadium redox flow battery based on situational awareness according to claim 3, characterized in that, In step 4, the electrolyte concentration method and the coulomb counting method are used to make a preliminary estimate of the SOC, and the results of the two methods are fused through an adaptive weighting factor. The core fusion formula is as follows: ;in An adaptive weighting factor that is dynamically adjusted based on sensor reliability. SOC calculated using the electrolyte concentration method. The SOC is calculated based on the Coulomb integral; the fusion results will be used. As the observation input, the extended Kalman filter achieves the optimal joint estimation of charge state, health state, polarization state and thermal management state; the fuzzy comprehensive evaluation method is adopted, and the optimal joint estimation is used as the evaluation factor. Through fuzzy synthesis operation and the maximum membership principle, the comprehensive operating status level of the system is identified.

5. The method for managing a vanadium redox flow battery based on situational awareness according to claim 4, characterized in that, For short-term forecasting, a long short-term memory network model with an attention mechanism is used to predict short-term grid demand and battery state trends. The attention mechanism focuses the model on key time-series information to predict short-term grid demand and battery state trends. For long-term forecasting, an improved Arrhenius model is used to predict battery life degradation, and an ARIMA-GARCH hybrid model is used to predict electricity prices and their fluctuations.

6. The method for managing a vanadium redox flow battery based on situational awareness according to claim 5, characterized in that, Step 6: Construct a multi-objective optimization function for model predictive control rolling, with the core form as follows: ;in 、 and As a weighting factor, 、 and These are sub-objective functions for efficiency, lifetime decay, and economic cost, respectively. In each control cycle, based on the function and the prediction model, the optimal control sequence that satisfies all safety constraints in the future time domain is solved, and the first term of the sequence is executed to achieve dynamic optimization of electrolyte flow rate, charging and discharging power, and thermal management.

7. The method for managing a vanadium redox flow battery based on situational awareness according to claim 6, characterized in that, Step 7 performs a security check on the optimization instructions; if the instructions... If a constraint is violated, the projective method is used to constrain it to the nearest safe boundary, as shown below: ;in For optimized quality, and To control the upper and lower limits of the safety of the variables, the safety instructions are converted into specific actuator signals through calibration relationships, and the execution deviation is monitored. When the threshold is exceeded, feedback adjustment is triggered.

8. The method for managing a vanadium redox flow battery based on situational awareness according to claim 7, characterized in that, Step 8 specifically includes: updating the parameters of the digital twin model online using a recursive least squares method with a forgetting factor λ. The core parameter update formula is: ;in, Here is the gain matrix. These are actual observed values. The observation vector is used to update the model parameters in real time to track the time-varying characteristics of the battery system, keeping the digital twin model synchronized with the physical entity.

9. A situational awareness-based management method for all-vanadium redox flow batteries according to claim 8, characterized in that, It also includes step 9: Based on the situation assessment and prediction results, automatically generate a system health status report and maintenance recommendations, and perform real-time data visualization, situation level display and multi-level alarm information push through the human-computer interaction interface; when the key performance indicators are lower than the preset threshold, the warning is automatically triggered in the maintenance recommendations.

10. A situational awareness-based all-vanadium redox flow battery management system, characterized in that, To implement the situational awareness-based management method for a vanadium redox flow battery as described in claim 9, the system comprises: a digital twin model management module, a multi-source data acquisition and fusion module, a situational awareness and prediction module, a predictive analysis unit, an intelligent decision-making and optimization module, a safety control and execution module, a model online update module, and a human-computer interaction and reporting module; the digital twin model management module is used to construct and maintain a multi-physics digital twin model of the vanadium redox flow battery, and to perform model initialization and parameter configuration; the multi-source data acquisition and fusion module is used to acquire real-time internal state data and external environmental data of the battery, and to perform data preprocessing and fusion; the situational awareness and prediction module is used to fuse electrolyte concentration method, coulomb counting method, and extended Kalman filter to estimate SOC, SOH, polarization state, and thermal management state in real time; and employs... The fuzzy comprehensive evaluation method assesses the system's operational status; the predictive analysis unit uses an attention-based LSTM model for short-term state prediction and combines an improved Arrhenius model and an ARIMA-GARCH model for long-term performance and market prediction; the intelligent decision-making and optimization module constructs multi-objective optimization functions and uses model predictive control algorithms to dynamically generate optimization strategies for electrolyte flow, charging and discharging power, thermal management, and economic scheduling; the safety control and execution module performs safety verification on optimization commands and converts them into control signals to drive actuators, while monitoring execution deviations; the online model update module uses a recursive least squares algorithm to update the digital twin model parameters in real time, ensuring synchronization between the model and the physical entity; and the human-computer interaction and reporting module visualizes real-time status data, generates health reports, issues maintenance suggestions, and manages alarm information.

Images