Intelligent operation and maintenance method for sodium ion battery energy management system based on digital twinning

Abstract

The invention is suitable for the technical field of sodium ion batteries, and provides a digital twinning-based intelligent operation and maintenance method for a sodium ion battery energy management system, and the method comprises the steps: constructing and operating a virtual model corresponding to a sodium ion battery system in a cloud digital twinning platform based on collected multi-source heterogeneous operation data; the virtual model simulates the internal state and evolution trend of the sodium ion battery system by coupling the electrochemical model, the thermal model and the life model; tracking and predicting the state of charge and the state of health of the sodium ion battery system, generating a charging and discharging power instruction and a thermal management strategy through a dynamic optimization control algorithm, and issuing and executing the instruction; and performing anomaly detection by applying an unsupervised learning algorithm, and generating predictive maintenance suggestions in combination with the prediction result of the health state. By constructing digital twinborn body and multi-model coupling, multi-dimensional accurate perception and multi-target dynamic optimization of the sodium ion battery system are realized, and the full life cycle value and safety level of the energy storage system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sodium ion battery, and particularly relates to an intelligent operation and maintenance method for a sodium ion battery energy management system based on digital twinning. BACKGROUND

[0002] At present, the operation and maintenance management of the sodium ion battery energy storage system is mainly based on the combined application of the battery management system and the monitoring and data acquisition system. Among them, the battery management system is responsible for realizing the basic battery protection, voltage and temperature monitoring, battery balancing and the estimation of key parameters such as state of charge and health state; the monitoring and data acquisition system is responsible for the collection, storage and visual display of operation data. In actual application, the charge and discharge management, thermal management and balancing strategy of the system are usually controlled based on the preset fixed parameter threshold, and the maintenance operation usually adopts the mode of regular inspection and post-failure maintenance.

[0003] However, the state estimation accuracy of the existing management system is insufficient, the existing ampere-hour integral method has cumulative error, the open circuit voltage method needs long-term static placement of the battery, and the equivalent circuit model is difficult to accurately describe the complex electrochemical process inside the battery, resulting in low reliability of the state of charge and health state estimation results, affecting the accuracy of system control. Secondly, the operation and maintenance strategy has poor flexibility, the charge and discharge and thermal management strategy based on fixed threshold cannot be dynamically adjusted according to the actual health state of the battery, environmental temperature and load change, resulting in low system energy efficiency and accelerating the battery life attenuation. In addition, regular maintenance and post-maintenance are difficult to identify early fault hidden dangers in advance, which is easy to make small faults evolve into serious accidents, increase downtime and maintenance cost.

[0004] In view of this, an intelligent operation and maintenance method for a sodium ion battery energy management system based on digital twinning is proposed. SUMMARY

[0005] The present application provides an intelligent operation and maintenance method for a sodium ion battery energy management system based on digital twinning, which solves the problem that the existing intelligent operation and maintenance method for a sodium ion battery energy management system cannot meet complex and high requirement operation and maintenance scenes.

[0006] The present application provides an intelligent operation and maintenance method for a sodium ion battery energy management system based on digital twinning, which includes: Collecting multi-source heterogeneous operation data of the sodium ion battery system and uploading to a cloud digital twinning platform; Based on the multi-source heterogeneous operating data, a virtual model corresponding to the sodium-ion battery system is constructed and run in the cloud-based digital twin platform. The virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. The electrochemical model is used to describe the ion concentration and potential distribution inside the battery, the thermal model is used to calculate the temperature field distribution of the battery under different operating conditions, and the lifetime model is used to predict the battery capacity decay and internal resistance growth. Based on the multi-source heterogeneous operating data and virtual model, the state of charge and health of the sodium-ion battery system are tracked and predicted using a state estimation algorithm. Based on the predicted state of charge and health, combined with external load demand and electricity price signals, a dynamic optimization control algorithm is used to generate charging and discharging power commands and thermal management strategies, which are then issued and executed. The system collects operational data generated after the control strategy is executed. Based on the deviation between historical operational data and the predicted state of the virtual model, an unsupervised learning algorithm is applied to detect anomalies. Predictive maintenance suggestions are then generated by combining the predicted health status results.

[0007] Furthermore, the electrochemical model is a pseudo-two-dimensional model that has undergone valence reduction processing. In this virtual model, the concentration and potential distribution of sodium ions inside the positive and negative electrode particles and in the electrolyte are described by the following set of equations: Solid-phase diffusion equation: Electrolyte phase mass conservation equation: Electrode reaction kinetic equation: In the formula: This refers to the concentration of lithium ions in the solid phase. The solid-phase diffusion coefficient is... For the radial coordinates of the particles, This refers to the lithium-ion concentration in the electrolyte phase. For time, This refers to the volume fraction of the electrolyte phase. The effective liquid phase diffusion coefficient, This represents the lithium-ion transference number. It is Faraday's constant. For local reaction current density, For exchange current density, and These are the anode and cathode transfer coefficients, respectively. This is an overpotential. The gas constant is This refers to absolute temperature.

[0008] Furthermore, the thermal model is a three-dimensional unsteady-state thermal model, in which the temperature field distribution is collected and distributed through the following coupling process: Based on the energy conservation equation: Among them, heat generation rate The electrochemical heat and Joule heat output by the electrochemical model are used to determine the following: In the formula: For density, For specific heat capacity, Thermal conductivity, For temperature, For time, Electrode The open circuit potential, For local current density, This is the solid-state potential. The calculated temperature field distribution, representing the liquid phase potential, is fed back to the electrochemical model in real time to correct the reaction rate constant, ionic conductivity, and diffusion coefficient.

[0009] Furthermore, the lifetime model is a degradation model based on the electrochemical and mechanical coupling mechanism. In the virtual model, battery degradation is predicted using the following coupling formula: The capacity decay rate is determined by the growth of the solid electrolyte interfacial film and the loss of active material: The internal resistance growth rate is expressed as: In the formula: and They are respectively Temperature-dependent rate constants for membrane growth and loss of active material The mechanical stress influence coefficient is... Mechanical stress, For the transmission coefficient, This is an overpotential. For electrode area, This is the initial maximum capacity. , , This is an empirical coefficient. The activation energy for internal resistance growth; The lifetime model receives the local current density from the electrochemical model. and the temperature of the thermal model As input, the calculated capacity and internal resistance changes are fed back to the electrochemical model to update the battery parameters.

[0010] Furthermore, the virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. The specific coupling process includes: The local heat generation rate calculated by the electrochemical model is used as a key input and passed to the thermal model to drive the temperature field simulation calculation. The temperature field distribution calculated by the thermal model is fed back to the electrochemical model and the lifetime model in real time to dynamically correct the reaction kinetic parameters and ion transport parameters in the electrochemical model and the decay rate parameters in the lifetime model. The capacity decay and internal resistance growth results predicted by the lifetime model based on the feedback temperature, local current density and state of charge are fed back to the electrochemical model to update its key state parameters, thereby realizing the simulation of the system state evolution trend under multi-physics coupling.

[0011] Furthermore, the step of tracking and predicting the state of charge and health of the sodium-ion battery system based on the multi-source heterogeneous operating data and virtual model using a state estimation algorithm includes: A state-space equation for a state vector is constructed based on the reduced-order electrochemical model in the virtual model. The collected real-time voltage and temperature data are used as observations to establish an observation equation with respect to the state vector. By using the unscented Kalman filter algorithm, the nonlinear characteristics of the state-space equation and the observation equation are processed through unscented transformation, and the state vector is recursively estimated to achieve real-time tracking of the state of charge and the state of health. Based on the state vector sequence obtained by recursive estimation, the long-term prediction of battery health status is achieved by extrapolating the changing trends of battery parameters through the lifetime model.

[0012] Furthermore, the method of using the unscented Kalman filter algorithm to recursively estimate the state vector by processing the nonlinear characteristics of the state-space equation and the observation equation through unscented transformation includes: Construct a state vector containing microscopic electrochemical states, wherein the state vector is represented as: in: and These represent the sodium ion concentrations on the surface of the positive and negative electrode particles, respectively. and These represent the average sodium ion concentrations of the positive and negative electrode particles, respectively. This represents the average sodium ion concentration in the electrolyte. In a charged state, For ohmic internal resistance, For charge transfer resistor, This represents the maximum usable capacity of the battery. The probability distribution of the state vector is approximated by a Sigma point sampling strategy, wherein the Sigma points are calculated from the mean and covariance of the state vector. Substituting the Sigma point into the state-space equation and performing nonlinear propagation, the predicted state mean and covariance are obtained. Substitute the predicted Sigma points into the observation equation to obtain the predicted observations; By calculating the Kalman gain, the predicted state and real-time observation data are fused to update the state vector estimate and covariance matrix; An adaptive mechanism is introduced to adjust the process noise covariance matrix and the observation noise covariance matrix in real time according to the innovation sequence in order to cope with model mismatch caused by battery aging.

[0013] Furthermore, the step of generating charging and discharging power commands and thermal management strategies based on the predicted state of charge and health, combined with external load demand and electricity price signals, and issuing and executing them through a dynamic optimization control algorithm includes: Construct a multi-objective optimization function that comprehensively considers grid interaction costs, battery degradation costs, and thermal management energy consumption costs: in: To predict the time domain, To control the input vector, the power grid interaction cost The calculation is as follows: Battery degradation cost The calculation is as follows: Thermal management energy consumption cost The calculation is as follows: In the formula: For grid interaction power, For real-time electricity prices, This represents the capacity degradation predicted by the lifetime model. The cost per unit capacity of battery replacement. For the power consumption of the thermal management system, This is the thermal management energy consumption cost coefficient. , , These are the weighting coefficients.

[0014] Furthermore, when solving the multi-objective optimization function, the dynamic optimization control algorithm applies constraints including voltage constraints, temperature constraints, current constraints, state of charge constraints, constraints to prevent sodium dendrite formation, constraints to prevent positive electrode phase transition, electrolyte concentration constraints, and temperature gradient constraints.

[0015] Furthermore, the predictive maintenance recommendations include at least one of the following: The command to perform active balancing maintenance on the specified battery cluster; To stabilize the solid electrolyte interface film, a small current repair charge / discharge command is executed. The module replacement warning is generated based on the health status prediction results, and the warning includes the expected replacement time. Instructions for cleaning or performance verification of specific components of the thermal management system; Discharge depth and charge rate adjustment recommendations based on the sodium dendrite formation risk index.

[0016] As can be seen from the above technical solutions, the present invention has the following advantages: This invention collects multi-source heterogeneous operating data of a sodium-ion battery system and uploads it to a cloud-based digital twin platform to construct and run a virtual model corresponding to the physical system. This model simulates the internal state and evolution trend of the battery system by coupling an electrochemical model, a thermal model, and a lifetime model. Based on the multi-source data and the virtual model, a state estimation algorithm is used to accurately track and predict the state of charge and health of the battery system. According to the prediction results, combined with external load demand and electricity price signals, a dynamic optimization control algorithm is used to generate optimal charging and discharging power commands and thermal management strategies and issue them for execution. New operating data generated after the execution of the control strategy is collected, and based on the deviation between historical data and the model's predicted state, an unsupervised learning algorithm is applied to detect anomalies. Combined with the health state prediction results, predictive maintenance suggestions are generated. This invention achieves precise perception of the internal microstate of sodium-ion batteries and accurate prediction of future trends by constructing a digital twin that integrates multi-physics mechanisms, thereby improving the estimation accuracy of state of charge and health. The dynamic optimization algorithm based on model predictive control can achieve comprehensive optimization of operating costs, battery degradation costs, and thermal management energy consumption while meeting multiple safety constraints, thus realizing a comprehensive improvement in the sodium-ion battery energy storage system in terms of operation and maintenance accuracy, economy, safety, and life cycle value. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of an embodiment of an intelligent operation and maintenance method for a sodium-ion battery energy management system based on digital twins in this invention. Figure 2 This is a schematic diagram of the coupling process of the electrochemical model, thermal model and lifetime model in this invention; Figure 3 This is a flowchart illustrating the process of predicting the state of charge and health of a battery in this invention. Detailed Implementation

[0018] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] Example 1 The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described below from the perspective of system implementation. Please refer to... Figure 1 The method provided in this application includes the following steps: S1. Collect multi-source heterogeneous operation data of the sodium-ion battery system and upload it to the cloud-based digital twin platform; In practice, a multi-type sensor network is deployed at the battery module and system levels to monitor the physical system's operating status in all dimensions. Specifically, voltage and current sensors collect the terminal voltage and loop current of each battery cell; multi-channel temperature sensors monitor the temperature of the battery surface, terminal connections, and key points inside the module with a resolution of 0.5 degrees Celsius; environmental sensors collect real-time data on ambient temperature, relative humidity, and atmospheric pressure within the battery compartment; and the thermal management subsystem reports real-time operating parameters such as coolant flow rate, water pump speed, and cooling fan status.

[0020] These multi-source, heterogeneous operational data are centrally collected and preprocessed through edge gateways deployed on-site. The edge gateways have built-in multi-channel analog-to-digital converters and signal conditioning circuits to ensure accurate acquisition of signals from different types of sensors. The collected data undergoes data cleaning, timestamp alignment, and protocol encapsulation to form standardized data packets, which are then uploaded in real-time to the cloud-based digital twin platform via industrial Ethernet or 5G mobile communication networks.

[0021] This cloud-based digital twin platform employs a microservice architecture, comprising a data access layer, a storage layer, and a computing layer. The data access layer receives data streams from multiple edge gateways in parallel through load balancing, and performs data validity verification and outlier filtering. The storage layer uses a hybrid architecture of time-series and relational databases to store frequently collected runtime data and system configuration parameters, respectively. The computing layer provides distributed computing resources to support subsequent virtual model construction and simulation analysis.

[0022] S2. Based on multi-source heterogeneous operating data, a virtual model corresponding to the sodium-ion battery system is constructed and run in a cloud-based digital twin platform. The virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. Among them, the electrochemical model is used to describe the ion concentration and potential distribution inside the battery, the thermal model is used to calculate the temperature field distribution of the battery under different operating conditions, and the lifetime model is used to predict the battery capacity decay and internal resistance growth. The virtual model here is a high-fidelity computer simulation system that integrates multiple physically meaningful sub-models to accurately reproduce the structure, behavior, and working rules of a physical battery system in digital space. It can also perform dynamic simulations based on real-time data, thereby simulating and predicting the internal state and future evolution trend of the physical system.

[0023] Specifically, the electrochemical model is a pseudo-two-dimensional model with reduced valence. In the virtual model, it is used to describe the concentration and potential distribution of sodium ions inside the positive and negative electrode particles and in the electrolyte through the following set of equations: Solid-phase diffusion equation: Electrolyte phase mass conservation equation: Electrode reaction kinetic equation: In the formula: This refers to the concentration of lithium ions in the solid phase. The solid-phase diffusion coefficient is... For the radial coordinates of the particles, This refers to the lithium-ion concentration in the electrolyte phase. For time, This refers to the volume fraction of the electrolyte phase. The effective liquid phase diffusion coefficient, This represents the lithium-ion transference number. It is Faraday's constant. For local reaction current density, For exchange current density, and These are the anode and cathode transfer coefficients, respectively. This is an overpotential. The gas constant is This refers to absolute temperature.

[0024] The model employs a pseudo-two-dimensional architecture with reduced order. The solid-phase diffusion equation describes the diffusion kinetics of sodium ions within the spherical particles of the positive and negative electrodes, where the rate of change of the solid-phase sodium ion concentration over time is determined by the solid-phase diffusion coefficient and the radial coordinate of the particles. The electrolyte phase mass conservation equation characterizes the migration process of sodium ions in the porous electrode electrolyte, which is influenced by the effective liquid-phase diffusion coefficient, the electrolyte phase volume fraction, and the lithium ion transference number. The electrode reaction kinetic equation adopts the Butler-Folmer form, quantifying the electrochemical reaction rate at the electrode-electrolyte interface through parameters such as exchange current density and anode-cathode transfer coefficients.

[0025] Specifically, the thermal model is a three-dimensional unsteady-state thermal model, and the temperature field distribution is collected and distributed in the virtual model through the following coupling process: Based on the energy conservation equation: Among them, heat generation rate The electrochemical heat and Joule heat output by the electrochemical model are used to determine the following: In the formula: For density, For specific heat capacity, Thermal conductivity, For temperature, For time, Electrode The open circuit potential, For local current density, This is the solid-state potential. The calculated temperature field distribution, representing the liquid phase potential, is fed back to the electrochemical model in real time to correct the reaction rate constant, ionic conductivity, and diffusion coefficient.

[0026] A three-dimensional unsteady-state thermal model is further integrated into the electrochemical model to simulate the temperature field distribution of the battery system. This thermal model is based on the law of conservation of energy, and its governing equations describe the relationship between the battery's internal temperature and time, which is jointly determined by thermal conduction and internal heat generation. The key heat generation rate term is directly provided by the electrochemical model, specifically including reversible heat caused by the entropy change of the electrochemical reaction and irreversible Joule heating caused by Ohmic and electrochemical polarization. The thermal model is discretized using a three-dimensional grid, enabling accurate simulation of the spatial temperature distribution of the battery under different cooling conditions.

[0027] Specifically, the lifetime model is a degradation model based on the electrochemical and mechanical coupling mechanism. In the virtual model, battery degradation is predicted using the following coupling formula: The capacity decay rate is determined by the growth of the solid electrolyte interfacial film and the loss of active material: The internal resistance growth rate is expressed as: In the formula: and They are respectively Temperature-dependent rate constants for membrane growth and loss of active material The mechanical stress influence coefficient is... Mechanical stress, For the transmission coefficient, This is an overpotential. For electrode area, This is the initial maximum capacity. , , This is an empirical coefficient. The activation energy for internal resistance growth; The lifetime model receives the local current density of the electrochemical model. and the temperature of the thermal model As input, the calculated capacity and internal resistance changes are fed back to the electrochemical model to update the battery parameters.

[0028] The lifetime model based on the electrochemical-mechanical coupling mechanism predicts capacity decay and internal resistance growth by quantifying two main decay mechanisms. The growth rate of the solid electrolyte interfacial film is significantly affected by temperature and overpotential, following an Arrhenius relation and exhibiting an exponential relationship with local current density. Active material loss is related to the square root term of the local current density, reflecting the irreversible phase transition and shedding of active material during cycling. Simultaneously, the model considers the accelerating effect of mechanical stress on battery decay, coupling the stress field with electrochemical decay through a mechanical stress influence coefficient. Internal resistance growth is modeled as a function of capacity decay, with growth components related to both current stress and mechanical stress considered separately.

[0029] Please see Figure 2 The virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. The specific coupling process includes: S21. The local heat generation rate calculated by the electrochemical model is used as the key input and passed to the thermal model to drive the temperature field simulation calculation. S22. The temperature field distribution calculated by the thermal model is fed back to the electrochemical model and lifetime model in real time to dynamically correct the reaction kinetic parameters, ion transport parameters in the electrochemical model and the decay rate parameters in the lifetime model. S23. The capacity decay and internal resistance growth results predicted by the lifetime model based on the feedback temperature, local current density and state of charge are fed back to the electrochemical model to update its key state parameters, thereby realizing the simulation of the system state evolution trend under multi-physics coupling.

[0030] Specifically, the local heat generation rate calculated by the electrochemical model at each simulation step is extracted in real time and passed to the thermal model as a key input parameter. Upon receiving this parameter, the thermal model maps it as a volumetric heat source term onto the corresponding 3D spatial mesh nodes, thereby driving the transient simulation calculation of the temperature field. The 3D temperature field distribution calculated by the thermal model is fed back to the electrochemical and lifetime models in real time. For the electrochemical model, this temperature field is used to dynamically correct its internal reaction kinetic parameters, including the reaction rate constant, ionic conductivity, and diffusion coefficient, all of which are functionally related to temperature through the Arrhenius equation. For the lifetime model, the temperature field is used to update the rate constants for solid electrolyte interfacial film growth and active material loss, enabling the degradation prediction to respond to the real-time thermal state of the battery. The lifetime model integrates the received temperature field, the local current density from the electrochemical model, and the state of charge to execute the degradation prediction algorithm. The calculated capacity degradation and internal resistance growth are fed back to the electrochemical model in real time to update its key state parameters, including maximum usable capacity and ohmic internal resistance. The above steps ensure that the electrochemical model can accurately reflect the performance degradation of the battery as it cycles and ages, thus achieving a high-fidelity simulation of the system state evolution trend under multi-physics coupling.

[0031] S3. Based on multi-source heterogeneous operating data and virtual models, the state of charge and health of sodium-ion battery systems are tracked and predicted using state estimation algorithms; This step employs a state estimation method based on the unscented Kalman filter algorithm. By fusing real-time monitoring data with mechanistic knowledge from a virtual model, it achieves high-precision prediction of the battery's state of charge and health. Please refer to [link / reference]. Figure 3 The specific implementation steps are as follows: S31. Construct the state-space equation of the state vector based on the reduced-order electrochemical model in the virtual model; State-space equations are mathematical models describing the evolution of a system's state over time. Based on the internal mechanism of the aforementioned reduced-order electrochemical model, a high-dimensional state vector is constructed that includes both microscopic electrochemical states and macroscopic electrical parameters. This state vector is expressed as: in: and These represent the sodium ion concentrations on the surface of the positive and negative electrode particles, respectively. and These represent the average sodium ion concentrations of the positive and negative electrode particles, respectively. This represents the average sodium ion concentration in the electrolyte. In a charged state, For ohmic internal resistance, For charge transfer resistor, This represents the maximum usable capacity of the battery. The discrete form of the state-space equation is expressed as: in: This represents the nonlinear state transition function described by the reduced-order electrochemical model; express The state vector at any given time; express The system input vector at any given time is mainly the operating current, measured in amperes, with positive for charging and negative for discharging; Represents the set of inherent parameters of the model; This represents process noise, used to characterize model uncertainties and unmodeled dynamics. The equation fully describes the internal states of the battery from... Time's up The dynamic evolution process at any given moment.

[0032] S32. Using the collected real-time voltage and temperature data as observations, establish observation equations with respect to the state vector; The observation equations establish the mathematical relationship between the internal state of the system and directly measurable external physical quantities. This invention uses the real-time acquired battery terminal voltage and battery surface temperature as observations, forming an observation vector. ;in, This indicates the battery terminal voltage, measured in volts. This indicates the surface temperature of the battery, expressed in degrees Celsius.

[0033] The general form of the observation equation is expressed as: in: Represents a nonlinear observation function. This represents observation noise and is used to characterize measurement error.

[0034] The specific observation equation for the terminal voltage is established based on an electrochemical model, and its expression is as follows: in: This represents the open-circuit potential of the positive electrode, which is a function of the sodium ion concentration on the surface of the positive electrode particles, and is measured in volts. This represents the open-circuit potential of the negative electrode, which is a function of the sodium ion concentration on the surface of the negative electrode particles, and is measured in volts. This indicates the operating current, measured in amperes. Represents the universal gas constant; This indicates the absolute temperature of the battery, measured in Kelvin. Denotes Faraday's constant; This represents the initial sodium ion concentration in the electrolyte, expressed in moles per cubic meter. This equation comprehensively reflects the battery's thermodynamic equilibrium potential and various overpotentials.

[0035] The specific equation for observing surface temperature is established based on a simplified thermal model, and its expression is: in: This indicates the ambient temperature, expressed in degrees Celsius. This represents the convective heat transfer coefficient, with units of watts per square meter per Kelvin; The value represents the battery surface area, expressed in square meters. This equation establishes the relationship between ohmic heat generation and surface temperature rise.

[0036] S33. Using the unscented Kalman filter algorithm, the nonlinear characteristics of the state space equation and the observation equation are processed through unscented transformation, and the state vector is recursively estimated to realize real-time tracking of the state of charge and the state of health. The unscented Kalman filter algorithm accurately processes nonlinear systems through unscented transformation. Its recursive estimation process includes the following six core steps: 1. Construct a state vector containing microscopic electrochemical states, represented as follows: 2. The probability distribution of the state vector is approximated by the Sigma point sampling strategy. The Sigma points are calculated from the mean and covariance of the state vector. exist At any given time, based on the current state estimate and its error covariance matrix Calculate a deterministic set of Sigma points. The calculation formula is as follows: in: The dimension of the state vector; This represents the scaling parameter, used to adjust the distance between the Sigma point and the mean; Representation matrix The square root matrix of the first List.

[0037] 3. Substitute the Sigma point into the state-space equation for nonlinear propagation to obtain the predicted state mean and covariance; Then calculate the mean of the predicted state. and prediction error covariance matrix .

[0038] 4. Substitute the predicted Sigma points into the observation equation to obtain the predicted observations; Calculate the mean of the predicted observations Covariance matrix of observations and the cross-covariance matrix of state and observation .

[0039] 5. By calculating the Kalman gain Integrating predicted state with real-time observation data Update the state vector estimate and covariance matrix; 6. An adaptive mechanism is introduced to adjust the process noise covariance matrix and the observation noise covariance matrix in real time according to the innovation sequence to cope with model mismatch caused by battery aging.

[0040] According to the new interest sequence The statistical characteristics are used to adjust the process noise covariance matrix in real time. and observation noise covariance matrix This enables the filter to adapt to model mismatch caused by factors such as battery aging, thereby enhancing the robustness of the algorithm and the long-term estimation accuracy.

[0041] S34. Based on the state vector sequence obtained by recursive estimation, the battery parameter change trend is extrapolated through the lifetime model to achieve long-term prediction of health status.

[0042] Long-term prediction is achieved by combining state estimation results with the aforementioned lifetime model. The system continuously records the state vector sequence obtained from recursive estimation, especially the maximum available capacity. With Ohm internal resistance Historical data. The time series of these key health status parameters were input into the lifespan model.

[0043] The lifetime model is based on its inherent electrochemical-mechanical coupling mechanism, and extrapolates the evolution trajectory of these parameters under future operating conditions using numerical integration methods. Specifically, this is achieved by solving the following coupled differential equations: in: This represents the rate constant for the growth of the solid electrolyte interfacial film. The rate constant representing the loss of active substances; Indicates the influence coefficient of mechanical stress; Indicates mechanical stress; Indicates the transmission coefficient; Indicates electrochemical overpotential; Indicates the electrode area; , , Represents the empirical coefficient; The activation energy representing the increase in internal resistance; This represents the local current density.

[0044] S4. Based on the predicted state of charge and health status, combined with external load demand and electricity price signals, generate charging and discharging power commands and thermal management strategies through dynamic optimization control algorithms and issue them for execution; To achieve the dual goals of economical operation and extended lifespan of sodium-ion battery energy storage systems, this step employs a model predictive control framework to construct a multi-objective optimization problem that comprehensively considers economic efficiency, battery health, and system safety. This optimization process is performed on a rolling basis in each control cycle, generating optimal charge / discharge power commands and thermal management strategies based on the latest system state predictions.

[0045] Construct a multi-objective optimization function that comprehensively considers grid interaction costs, battery degradation costs, and thermal management energy consumption costs: The objective function of the optimization problem aims to minimize the total operating cost in the prediction time domain, and its mathematical expression is: in: To predict the time domain, To control the input vector, the power grid interaction cost The calculation is as follows: Battery degradation cost The calculation is as follows: Thermal management energy consumption cost The calculation is as follows: In the formula: For grid interaction power, For real-time electricity prices, This represents the capacity degradation predicted by the lifetime model. The cost per unit capacity of battery replacement. For the power consumption of the thermal management system, This is the thermal management energy consumption cost coefficient. , , These are the weighting coefficients.

[0046] The dynamic optimization control algorithm applies the following constraints when solving the multi-objective optimization function: voltage constraint, temperature constraint, current constraint, state of charge constraint, constraint to prevent sodium dendrite formation, constraint to prevent positive electrode phase transition, electrolyte concentration constraint, and temperature gradient constraint. The expressions are as follows: To ensure the safe operation of the system and prevent specific aging mechanisms, the optimization problem must satisfy the following multiple constraints: The system operates under normal constraints to ensure it runs within a safe operating range: Voltage constraint: ,in This refers to the battery terminal voltage. and These are its allowed lower and upper limits, respectively.

[0047] Temperature constraints: ,in For battery temperature, and These are its allowed lower and upper limits, respectively.

[0048] Current constraint: ,in This is the battery operating current. and These are its allowed lower and upper limits, respectively.

[0049] State of charge constraints: ,in The battery is in its state of charge. and These are its allowed lower and upper limits, respectively.

[0050] Special safety constraints based on electrochemical models prevent specific faults at the mechanistic level: Constraints to prevent sodium dendrite formation: ;in, This represents the sodium ion concentration on the surface of the negative electrode particles. The minimum concentration required to prevent sodium deposition; The local current density at the negative electrode. This is a temperature-dependent critical current density; exceeding this value significantly increases the risk of dendrite formation.

[0051] Preventing positive electrode phase transition constraint: ;in, The rate of change of the state of charge. It is its maximum allowed value; and This is a known charge state range that can easily trigger harmful phase transitions in cathode materials, and must be avoided during operation.

[0052] Electrolyte concentration constraints: ;in, This represents the sodium ion concentration in the electrolyte. and These are the lower and upper limits that are allowed, respectively, to prevent excessively low concentrations from exacerbating side reactions or excessively high concentrations from causing mass transfer difficulties.

[0053] Temperature gradient constraint: ;in, The temperature gradient inside the battery. To set its maximum allowable value, in order to control thermal stress and avoid damage to the battery structure.

[0054] By solving the constrained multi-objective optimization problem described above, the optimal control sequence can be obtained. The first control variable in the sequence, namely the current charging / discharging power command and thermal management strategy, is sent to the battery management system and thermal management controller for execution, thereby achieving the optimal balance between energy efficiency and lifespan.

[0055] S5. Collect the operational data generated after the execution of the control strategy. Based on the deviation between historical operational data and the predicted state of the virtual model, apply an unsupervised learning algorithm to detect anomalies and generate predictive maintenance suggestions by combining the prediction results of the health status.

[0056] To achieve a shift from reactive to predictive maintenance, this invention establishes a complete anomaly detection and maintenance decision generation process. This process continuously monitors the discrepancy between the system's actual behavior and the expected behavior of its digital twin, identifies potential faults early, and, combined with a lifespan prediction model, generates forward-looking maintenance recommendations.

[0057] The system continuously collects new operational data generated after executing optimized control strategies, including key parameters such as voltage, current, and temperature. This real-time data, along with historical operational data, forms the basis of the analysis. By comparing actual measured values ​​with the predicted values ​​of the virtual model, a residual sequence is calculated. These residuals reflect the deviation between the actual operating state of the system and the expectations of the ideal model, serving as important indicators for identifying anomalies. An unsupervised learning algorithm, Isolation Forest, is employed for anomaly detection. First, a comprehensive feature vector with eighteen dimensions is constructed. This vector comprehensively characterizes the battery's performance state, including voltage consistency characteristics, temperature distribution characteristics, internal resistance and impedance characteristics, capacity and energy efficiency characteristics, as well as sodium-ion battery-specific characteristics, such as the rate of change of sodium ion diffusion coefficient, the rate of increase of solid electrolyte interface film impedance, the positive electrode phase transition risk index, and the sodium dendrite formation risk index.

[0058] The Isolation Forest algorithm isolates outlier data points by constructing multiple random decision trees. For each feature vector... The algorithm calculates its anomaly score. The formula is: in: Representing the eigenvector The average path length in a multi-isolated tree, which is the number of edges traversed from the root node to the leaf node; Given sample size The normalization factor at that time is used to standardize the path length; The sample size is represented by the outlier score. The closer the outlier score is to 1, the more likely the data point is to be an outlier.

[0059] To improve the reliability of detection results, anomaly detection results are cross-validated with a lifetime model based on electrochemical and mechanical coupling. A joint anomaly index is defined, which combines the anomaly score calculated by the isolated forest algorithm and the model prediction bias, and the final evaluation result is obtained through weighted summation. When short-term anomaly detection and long-term anomaly trends occur simultaneously, the system increases the warning confidence level to ensure the accuracy of the warning.

[0060] Based on the results of cross-validation between anomaly characteristics and lifetime models, the system performs root cause analysis and generates targeted predictive maintenance recommendations. Specific maintenance recommendations include the following types: The instruction to perform active balancing maintenance on a specified battery cluster is generated when the voltage or state of charge inconsistency between individual cells within the battery cluster exceeds a threshold. The command to perform small-current repair charge and discharge is executed to stabilize the solid electrolyte interface membrane. It is triggered when the impedance of the solid electrolyte interface membrane increases abnormally and the capacity decay accelerates. The module replacement warning is generated based on the health status prediction results. It is automatically generated when the predicted remaining service life is lower than a preset threshold. The warning information includes the specific expected replacement time. Cleaning or performance verification commands for specific components of the thermal management system are issued when a decrease in cooling efficiency or abnormal temperature distribution is detected. Based on the discharge depth and charge rate adjustment suggestions generated by the sodium dendrite formation risk index, an optimized operation strategy is proposed when the sodium ion concentration on the negative electrode surface is close to the critical value or the local current density is too high.

[0061] The above embodiments, by constructing a high-fidelity digital twin and a multi-model coupling mechanism, achieve multi-dimensional accurate perception and multi-objective dynamic optimization of sodium-ion battery systems from microscopic state to macroscopic performance, effectively improving the full life cycle value, operational economy and intrinsic safety level of energy storage systems.

[0062] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0063] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart operation and maintenance method for a sodium-ion battery energy management system based on digital twins, characterized in that, include: Collect multi-source heterogeneous operation data of sodium-ion battery systems and upload them to a cloud-based digital twin platform; Based on the multi-source heterogeneous operating data, a virtual model corresponding to the sodium-ion battery system is constructed and run in the cloud-based digital twin platform. The virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. The electrochemical model is used to describe the ion concentration and potential distribution inside the battery, the thermal model is used to calculate the temperature field distribution of the battery under different operating conditions, and the lifetime model is used to predict the battery capacity decay and internal resistance growth. Based on the multi-source heterogeneous operating data and virtual model, the state of charge and health of the sodium-ion battery system are tracked and predicted using a state estimation algorithm. Based on the predicted state of charge and health, combined with external load demand and electricity price signals, a dynamic optimization control algorithm is used to generate charging and discharging power commands and thermal management strategies, which are then issued and executed. The system collects operational data generated after the control strategy is executed. Based on the deviation between historical operational data and the predicted state of the virtual model, an unsupervised learning algorithm is applied to detect anomalies. Predictive maintenance suggestions are then generated by combining the prediction results of the health status.

2. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 1, characterized in that, The electrochemical model is a pseudo-two-dimensional model that has undergone valence reduction processing. In this virtual model, the concentration and potential distribution of sodium ions inside the positive and negative electrode particles and in the electrolyte are described by the following set of equations: Solid-phase diffusion equation: Electrolyte phase mass conservation equation: Electrode reaction kinetic equation: In the formula: This refers to the concentration of lithium ions in the solid phase. The solid-phase diffusion coefficient is... For the radial coordinates of the particles, This refers to the lithium-ion concentration in the electrolyte phase. For time, This refers to the volume fraction of the electrolyte phase. The effective liquid phase diffusion coefficient, This represents the lithium-ion transference number. It is Faraday's constant. For local reaction current density, For exchange current density, and These are the anode and cathode transfer coefficients, respectively. This is an overpotential. The gas constant is This refers to absolute temperature.

3. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 2, characterized in that, The thermal model is a three-dimensional unsteady-state thermal model, and the temperature field distribution in the virtual model is collected and distributed through the following coupling process: Based on the energy conservation equation: Among them, heat generation rate The electrochemical heat and Joule heat output by the electrochemical model are used to determine the following: In the formula: For density, For specific heat capacity, Thermal conductivity, For temperature, For time, Electrode The open circuit potential, For local current density, This is the solid-state potential. The calculated temperature field distribution, representing the liquid phase potential, is fed back to the electrochemical model in real time to correct the reaction rate constant, ionic conductivity, and diffusion coefficient.

4. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 3, characterized in that, The lifetime model is a degradation model based on the electrochemical and mechanical coupling mechanism. In the virtual model, battery degradation is predicted using the following coupling formula: The capacity decay rate is determined by the growth of the solid electrolyte interfacial film and the loss of active material: The internal resistance growth rate is expressed as: In the formula: and They are respectively Temperature-dependent rate constants for membrane growth and loss of active material The mechanical stress influence coefficient is... Mechanical stress, For the transmission coefficient, This is an overpotential. For electrode area, This is the initial maximum capacity. , , This is an empirical coefficient. The activation energy for internal resistance growth; The lifetime model receives the local current density from the electrochemical model. and the temperature of the thermal model As input, the calculated capacity and internal resistance changes are fed back to the electrochemical model to update the battery parameters.

5. The intelligent operation and maintenance method for a sodium-ion battery energy management system based on digital twins according to any one of claims 1 to 4, characterized in that, The virtual model simulates the internal state and evolution trend of the sodium-ion battery system by coupling an electrochemical model, a thermal model, and a lifetime model. The specific coupling process includes: The local heat generation rate calculated by the electrochemical model is used as a key input and passed to the thermal model to drive the temperature field simulation calculation. The temperature field distribution calculated by the thermal model is fed back to the electrochemical model and the lifetime model in real time to dynamically correct the reaction kinetic parameters and ion transport parameters in the electrochemical model and the decay rate parameters in the lifetime model. The capacity decay and internal resistance growth results predicted by the lifetime model based on the feedback temperature, local current density and state of charge are fed back to the electrochemical model to update its key state parameters, thereby realizing the simulation of the system state evolution trend under multi-physics coupling.

6. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 1, characterized in that, The step of tracking and predicting the state of charge and health of the sodium-ion battery system based on the multi-source heterogeneous operating data and virtual model using a state estimation algorithm includes: A state-space equation for a state vector is constructed based on the reduced-order electrochemical model in the virtual model. The collected real-time voltage and temperature data are used as observations to establish an observation equation with respect to the state vector. By using the unscented Kalman filter algorithm, the nonlinear characteristics of the state-space equation and the observation equation are processed through unscented transformation, and the state vector is recursively estimated to achieve real-time tracking of the state of charge and the state of health. Based on the state vector sequence obtained by recursive estimation, the long-term prediction of battery health status is achieved by extrapolating the changing trends of battery parameters through the lifetime model.

7. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 6, characterized in that, The method of using the unscented Kalman filter algorithm to recursively estimate the state vector by processing the nonlinear characteristics of the state-space equation and the observation equation through unscented transformation includes: Construct a state vector containing microscopic electrochemical states, wherein the state vector is represented as: in: and These represent the sodium ion concentrations on the surface of the positive and negative electrode particles, respectively. and These represent the average sodium ion concentrations of the positive and negative electrode particles, respectively. This represents the average sodium ion concentration in the electrolyte. In a charged state, For ohmic internal resistance, For charge transfer resistor, This represents the maximum usable capacity of the battery. The probability distribution of the state vector is approximated by a Sigma point sampling strategy, wherein the Sigma points are calculated from the mean and covariance of the state vector. Substituting the Sigma point into the state-space equation and performing nonlinear propagation, the predicted state mean and covariance are obtained. Substitute the predicted Sigma points into the observation equation to obtain the predicted observations; By calculating the Kalman gain, the predicted state and real-time observation data are fused to update the state vector estimate and covariance matrix; An adaptive mechanism is introduced to adjust the process noise covariance matrix and the observation noise covariance matrix in real time according to the innovation sequence in order to cope with model mismatch caused by battery aging.

8. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 1, characterized in that, The step of generating charging and discharging power commands and thermal management strategies based on the predicted state of charge and health, combined with external load demand and electricity price signals, and issuing them for execution through a dynamic optimization control algorithm includes: Construct a multi-objective optimization function that comprehensively considers grid interaction costs, battery degradation costs, and thermal management energy consumption costs: in: To predict the time domain, To control the input vector, the power grid interaction cost The calculation is as follows: Battery degradation cost The calculation is as follows: Thermal management energy consumption cost The calculation is as follows: In the formula: For grid interaction power, For real-time electricity prices, This represents the capacity degradation predicted by the lifetime model. The cost per unit capacity of battery replacement. For the power consumption of the thermal management system, This is the thermal management energy consumption cost coefficient. , , These are the weighting coefficients.

9. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin as described in claim 8, characterized in that, When solving the multi-objective optimization function, the dynamic optimization control algorithm applies constraints including voltage constraints, temperature constraints, current constraints, state of charge constraints, constraints to prevent sodium dendrite formation, constraints to prevent positive electrode phase transition, electrolyte concentration constraints, and temperature gradient constraints.

10. The intelligent operation and maintenance method for sodium-ion battery energy management system based on digital twin according to claim 1, characterized in that, The predictive maintenance recommendations include at least one of the following: The command to perform active balancing maintenance on the specified battery cluster; To stabilize the solid electrolyte interface film, a small current repair charge / discharge command is executed. The module replacement warning is generated based on the health status prediction results, and the warning includes the expected replacement time. Instructions for cleaning or performance verification of specific components of the thermal management system; Discharge depth and charge rate adjustment recommendations based on the sodium dendrite formation risk index.

Images