A dynamic power state based model predictive control charging optimization method

Abstract

The application relates to a dynamic power state-based model predictive control charging optimization method, a dynamic power state collaborative optimization mechanism is created, a real-time power upper limit is taken as an active optimization target instead of a fixed constraint condition, and a technical bottleneck that power limitation is passively responded in a traditional charging strategy is broken through; a specific method is that a lithium ion battery electric-thermal-aging multi-physical field coupling model is constructed, electric-thermal characteristic parameters are updated in real time through an online parameter identification algorithm, and a core temperature and an aging state are synchronously estimated through a double Kalman filter state observer; a four-dimensional target function optimization model containing a dynamic power state is innovatively established, a power upper limit, a charging speed, a capacity attenuation rate and a current fluctuation rate are subjected to multi-target collaborative optimization; a dynamic rolling optimization algorithm is designed based on a model predictive control framework, and the optimal charging current meeting the dynamic power distribution demand of a power grid is solved in real time under the hard constraint of guaranteeing the maximum core temperature and the terminal voltage.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of lithium battery charging technology, in particular to a model predictive control charging optimization method based on a dynamic power state. BACKGROUND

[0002] With the transformation of global energy structure, electric vehicles, as the core carrier of clean transportation, are accelerating the replacement of traditional fuel vehicles. However, electric vehicles have been facing two major challenges: high energy cost (mainly from the battery system) and long charging time. Although the price of batteries is expected to decrease, prolonging the service life of the battery is an effective means to reduce the cost at present. The aging of the battery is mainly affected by the charging strategy, and there is often a trade-off between the charging speed and the service life of the battery. Therefore, in order to charge quickly while minimizing the impact on the service life of the battery, it is essential to optimize the charging strategy.

[0003] In order to solve this problem, the breakthrough of charging technology has become the key to the large-scale commercial application of electric vehicles. As the most original method, the constant current and constant voltage method is widely used in many electric vehicles. However, these methods are essentially static and have limited understanding of the dynamics of the battery, and cannot accurately control the factors that affect the safety and aging rate of the battery. In contrast, model-based charging methods have shown significant advantages in battery system simulation, parameter optimization, state estimation and closed-loop control, and are therefore more suitable for high-precision and adaptive battery charging management.

[0004] Developing model-based battery charging algorithms faces two major challenges: first, the complex nonlinear dynamics of lithium-ion batteries with multiple physical fields (electric, electrochemical, thermal, and aging) coupled, with high computational cost of their multi-time / space scale partial differential equation models, making real-time application difficult; second, key state parameters (such as state of charge reflecting remaining capacity and state of health representing health degree) rely on internal variables (such as ion concentration and capacity decay) that cannot be directly measured, forcing researchers to combine model simplification and state estimation techniques to achieve a feasible solution. Existing model-based charging strategy research mostly targets voltage and temperature-constrained models and does not consider dynamic power and aging rate constraints. In summary, there is still a lack of an online application of an electric vehicle fast charging method that takes into account the speed, safety and life expectancy. SUMMARY

[0005] In order to solve the above technical problems, the application provides a lithium battery fast charging method based on a model predictive control method, which gives the constraint conditions for actual application according to the actual physical characteristic equation of the battery model, and ensures efficient and fast charging behavior of the battery.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for fast charging a lithium battery based on dynamic power state, comprising the following steps:

[0007] S1. Establish an electro-thermal-aging coupling model for lithium batteries. And the electro-thermal-aging coupled model Discretization is performed to obtain the discretized electro-thermal-aging coupled model. ;

[0008] S2. Conduct pulse charge-discharge experiments on lithium batteries to obtain real-time terminal voltage and surface temperature data, and input the collected terminal voltage and surface temperature data into the discretized electro-thermal-aging coupled model. ;

[0009] S3. Design a dual Kalman filter state observer for the dispersed electro-thermal-aging coupled model. Observations were conducted, and the discretized electro-thermal-aging coupling model was used. Simultaneous estimation yields the battery state of charge and core temperature-aging joint state variables;

[0010] S4. Conduct constant current and constant voltage charging experiments on lithium batteries, construct a four-dimensional objective function with charging current as the control variable, considering charging time, capacity decay rate, current fluctuation rate and power deviation, and combine the physical constraints and safety constraints of the battery to form a multi-constraint optimization problem.

[0011] S5. Establish a power state model and calculate the power state and power deviation. Construct a weight matrix for a four-dimensional objective function based on the power state, power deviation, state of charge, and core temperature-aging joint state variables.

[0012] S6. Use a quadratic programming solver to solve the multi-constraint optimization problem of the model at the current time step and generate the optimal charging sequence.

[0013] Furthermore, the electro-thermal-aging coupling model of the lithium battery in S1 includes an electrical model, which adopts a first-order RC equivalent circuit model. The expression of the circuit model is as follows:

[0014] ;

[0015] ;

[0016] in, This represents the derivative of the state of charge with respect to time. It refers to the battery's state of charge. It is the battery's input current. This represents the derivative of the polarization voltage with respect to time. Indicates polarization voltage. Indicates the battery's rated capacity. and These represent the polarization resistor and polarization capacitor, respectively.

[0017] Furthermore, the electro-thermal-aging coupling model of the lithium battery in S1 also includes a lumped parameter thermal model, where the rate of change of battery temperature is determined by the relationship between heat generation and heat dissipation, and its expression is as follows:

[0018] ;

[0019] in, This represents the derivative of temperature with respect to time. It is a heat-generating component of the battery. It is the ambient temperature. It's the battery temperature; It's thermal resistance. It is heat capacity.

[0020] Furthermore, the electro-thermal-aging coupling model of the lithium battery in S1 also includes a semi-empirical aging model, expressed as follows:

[0021] ;

[0022] ;

[0023] in It refers to the battery charge / discharge rate. It is the capacity decay, and yes The pre-exponential factor of the multiplication function, It is activation energy. It is the basic number of aging rate. It is an index of the sensitivity of aging rate to dynamic scaling factor. It is the time step Instantaneous current, It is the universal gas constant. It is the absolute temperature of the battery.

[0024] Furthermore, the S5 power state model is a multi-constraint joint model that considers voltage, temperature, and aging, and its expression is as follows:

[0025] ;

[0026] in, This is the upper limit of voltage. Open circuit voltage, The upper limit of the core temperature, It is the core temperature. For thermal resistance, It is the battery current. It is the capacity decay threshold. It is the capacity decay value. It is the battery's internal resistance in ohms. and It is an aging factor.

[0027] Furthermore, the discretization process in S1 employs a recursive least squares method with a forgetting factor to identify the electrical model parameters and the lumped parameter thermal model parameters online. The specific method is as follows:

[0028] The expression obtained after discretizing the circuit model is as follows:

[0029] ;

[0030] in, and Let be the terminal voltage and current at time k, respectively. Sampling time, The ohmic internal resistance of the battery is mainly composed of the intrinsic resistance of the electrolyte and the electrodes. and They jointly characterize the polarization effect and describe the dynamic behavior of the battery during charging and discharging.

[0031] Furthermore, the expression obtained after discretizing the lumped parameter thermal model is as follows:

[0032] ;

[0033] in and It is thermal resistance and heat capacity. , yes The ambient temperature at any given time yes Battery temperature at any time yes The amount of heat generated at any given moment.

[0034] Furthermore, the charged state of the S3 electrical model is used as the system's state variable, and the unscented Kalman-Busch algorithm is used to construct the state observer. The state equations and observation equations are as follows:

[0035] ;

[0036] ;

[0037] in, The SOC prediction value at the current moment. Let be the state transition function given by the equivalent circuit model. This is the posterior estimate of SOC from the previous time step. is the known input at the last time, is the actual measurement at the current time, is the output function given by the model, which maps SOC to terminal voltage, is the known input at the current time, is the and is the process noise and measurement noise.

[0038] Further, the physical constraints and safety constraints of the S4 battery are as follows:

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] wherein, is the maximum charge and discharge current, and are the upper and lower limits of the battery voltage, and are the upper and lower limits of the battery core temperature, is the maximum single-step capacity attenuation amount, is the nominal capacity.

[0044] Further, the four-dimensional target function expression in S4 is as follows:

[0045] ;

[0046] wherein, the symbol "hat" represents the estimated value, is the target SOC value, and respectively represent the target input current and the capacity loss term, and N is the prediction time domain, is the weighted square 2-norm, is the real-time charging power, is the power state upper limit P, Q, R, S at the current time, and P, Q, R, S are the weight matrix to be adjusted. Compared with the prior art, the technical scheme of the present application has the following beneficial effects:

[0047]

[0048] ​1、The dynamic power state-based model predictive control charging optimization strategy identifies battery internal resistance, polarization capacitance and other parameters in real time through recursive least squares with a forgetting factor, synchronously fuses power grid load rate, energy storage SOC and other external constraints, and establishes a real-time correction model of power upper limit. During the peak period of the power grid, i.e. when the load rate is > 90%, the system automatically adjusts the power upper limit to 60%-70% of the rated value, and through a power deviation penalty term with a weight coefficient δ = 1.8, controls the charging power fluctuation within ±8%, ensuring the stability of the power grid; during the valley period, the power upper limit is dynamically increased to 110%-120% of the rated value, combined with a current fluctuation rate suppression strategy with a weight coefficient γ = 0.6, the power utilization rate is increased from 75% of the traditional strategy to more than 92%. The actual measurement of a 30kWh power battery pack shows that after adopting the mechanism, the power grid interaction compatibility is improved by 40%, and the charging efficiency difference between peak and valley periods is reduced to within 15%.

[0049] 2、The dynamic power state-based model predictive control charging optimization strategy, based on the core temperature estimation of the lumped parameter thermal model and the unscented Kalman filter, constructs a power-temperature dual variable constraint mechanism: when the core temperature exceeds 45℃, the power upper limit decreases linearly with the temperature, i.e. the power is reduced by 3% for every 1℃ increase, and at the same time the temperature constraint weight is activated for adaptive adjustment, increasing the weight coefficient β from 0.5 to 1.2, and suppressing heat generation by reducing the charging current. This mechanism reduces the core temperature of the battery by 8-10℃ under 2C charging conditions compared with the traditional strategy, and reduces the critical risk probability of thermal runaway by 70%. In addition, the online identification of power-thermal resistance parameters has an identification error of <3%, supporting dynamic adaptation of the power boundary under different ambient temperatures, and maintaining a charging efficiency fluctuation of ≤12% within a temperature range of -10℃ to 50℃.

[0050] 3、Innovatively, the dynamic power upper limit, charging time, capacity attenuation rate and current fluctuation rate are included in the unified optimization target, and the multi-objective coordination is realized through adaptive adjustment of the weight matrix. In the fast charging mode, i.e. when the target SOC is from 20% to 80%, the system automatically increases the charging speed weight, at this time α = 1.5, and through the capacity attenuation penalty term β = 0.8 calculated by the aging model, the single-step attenuation is controlled within 0.01% Qnom, which shortens the charging time by 25% compared with the CCCV strategy, while prolonging the cycle life by 18%-22%. When the battery state of health (SOH) is lower than 85%, the aging constraint weight is automatically increased by 30%, and the power upper limit is reduced by 15%-20% to suppress irreversible capacity loss. Experimental verification shows that this strategy can reduce the aging rate by 28%.

[0051] 4, Establish a semi-empirical model of power change rate and SEI film growth rate, embed a power change penalty term in the objective function, and the weight coefficient η=0.4. When the power mutation amplitude exceeds 20% of the rated value, the current smoothing control strategy is automatically triggered. This mechanism can control the power change rate within 5W / s, effectively reduce the lithium deposition side reaction, and reduce the SEI film thickness growth rate by 35%. Through ternary lithium battery cycle test, the capacity retention rate after 1000 cycles is improved from 72% of the traditional strategy to 81%, and the internal resistance increase is reduced by 22%.

[0052] 5, By using model discretization and linear approximation technology, the nonlinear problem of electric-thermal-aging coupling is transformed into a quadratic programming model, and the control time domain is realized through parallel computing architecture. The optimal current sequence is generated within N=10-15, and the single solving time is <20ms, which meets the real-time control demand. When the power limit mutation ±30% is detected, the system adjusts the charging plan in advance by 2 control periods through the feedforward compensation mechanism, so that the power tracking error is <5%, and the response speed is improved by 3 times compared with the traditional feedback control strategy.

[0053] 6, The SOC, core temperature, aging state and power grid power margin are estimated simultaneously by using double Kalman filter algorithm, and the state estimation error is <1.5% SOC and 2℃, which provides high reliability input for power boundary generation. When the battery parameters drift, such as internal resistance aging growth of 20%, the recursive least squares method with forgetting factor can complete parameter update within 5-10 sampling periods, ensure the prediction accuracy of the model, and avoid power overshoot caused by model mismatch. Experimental verification shows that this mechanism can keep the charging strategy effective during battery aging, and the charging efficiency decay is ≤8% when SOH≤70%.

[0054] 7, Build a power-temperature-voltage multi-protection threshold system: when the power approaches the upper limit and the voltage reaches 4.15V, automatically start the "power-voltage" dual derating strategy, limit the current to within 1.2C; if the core temperature exceeds 50℃, activate the temperature priority mode, then the weight coefficient β=2.0, force the power to be reduced to less than 50% of the rated value. This mechanism can still run safely under the conditions of battery single voltage difference >50mV, sudden change of ambient temperature, i.e. ±10℃ / min, etc. The charging interruption probability is reduced by more than 95% compared with the traditional strategy. In extreme condition test, the system can quickly identify and execute partition power regulation when the battery monomer overheats, i.e. local temperature 60℃, avoid fault diffusion, and ensure the safety of the battery. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 is the flowchart of the lithium battery optimization fast charging method based on model predictive control of the present application

[0056] Figure 2 Figure of the electro-thermal coupling model of the cylindrical lithium battery of the present application;

[0057] Figure 3 Results of parameter identification using the least square method with a forgetting factor of the present application;

[0058] Figure 4 Figure of voltage effect comparison between the present application and the traditional constant current constant voltage charging algorithm;

[0059] Figure 5 Figure of temperature effect comparison between the present application and the traditional constant current constant voltage charging algorithm;

[0060] Figure 6 Figure of state of charge effect comparison between the present application and the traditional constant current constant voltage charging algorithm;

[0061] Figure 7 Figure of current effect comparison between the present application and the traditional constant current constant voltage charging algorithm. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0063] Please refer to Figures 1-7 , in the present example, a model predictive control charging optimization method based on dynamic power state is provided, comprising:

[0064] S1, establishing an electro-thermal-aging coupling model of a lithium battery , and performing discretization processing on the electro-thermal-aging coupling model to obtain a discretized electro-thermal-aging coupling model ;

[0065] The present application adopts an electro-thermal coupling model as Figure 1 shown to describe the characteristics of a lithium ion battery, and the electrical model expression is as follows:

[0066] ;

[0067] ;

[0068] wherein, represents the derivative of the state of charge with respect to time, is the state of charge of the battery, is the input current of the battery, denotes the derivative of the polarization voltage with respect to time, denotes the polarization voltage, denotes the rated capacity of the battery, and denotes the polarization resistance and the polarization capacitance, respectively.

[0069] The terminal voltage of the battery is expressed as follows:

[0070]

[0071] wherein is the open-circuit voltage of the battery, is the ohmic internal resistance of the battery.

[0072] The thermal model built is a lumped parameter thermal model, and the rate of change of the battery temperature is determined by the heat generation and the heat dissipation, which is expressed as follows:

[0073]

[0074] wherein denotes the derivative of the temperature with respect to time, is the heat generation term of the battery, is the ambient temperature, is the battery temperature; is the thermal resistance, is the thermal capacity.

[0075] The expression of the heat generation term is as follows:

[0076]

[0077] wherein is the partial derivative of the open-circuit voltage of the battery with respect to the battery temperature.

[0078] The aging model adopted is a semi-empirical aging model, which is expressed as follows:

[0079]

[0080]

[0081] wherein is the battery charge-discharge rate, is the capacity attenuation amount, and is the pre-factor of the rate function, the power-law factor z = 0.55, is the basic number of the aging rate, is the sensitivity index of the aging rate to the dynamic rate, is the instantaneous current of the time step ​​​​​​is the universal gas constant, is the absolute temperature of the battery. is the activation energy calculated from the following formula:

[0082] ;

[0083] The State of Power (SOP) model is a multi-constraint joint model considering voltage, temperature, and aging, and its expression is as follows:

[0084] ;

[0085] wherein, is the upper limit of the voltage, is the open circuit voltage, is the upper limit of the core temperature, is the core temperature, is the thermal resistance, is the battery current, is the capacity decay threshold, is the capacity decay value, is the battery ohmic internal resistance, and are the aging factors.

[0086] Further, the recursive least squares method with a forgetting factor is used to identify the parameters of the electro-thermal model;

[0087] First, based on real battery data collection, the parameters of the electro-thermal model are identified using the recursive least squares method with a forgetting factor (FFRLS, Forgetting Recursive Least Squares), and then the unscented Kalman-Bucy filter (UKBF, Unscented Kalman-Bucy Filter) is used to estimate the state of charge in real time.

[0088] The expression obtained after discretizing the electrical model is as follows:

[0089] ;

[0090] wherein, and are the terminal voltage and current at time k, is the sampling time, is the ohmic internal resistance of the battery, mainly composed of the intrinsic resistance of the electrolyte and the electrode; and together represent the polarization effect and describe the dynamic behavior of the battery during charging and discharging.

[0091] The regression vector and the parameter vector are as follows:

[0092] ;

[0093] ;

[0094] where the regression vector is obtained from the historical observation data of the input model, and the parameter vector is obtained from the unknown parameters to be identified in the model.

[0095] The electrical identification model is as follows:

[0096] ;

[0097] where, is the output voltage of the model at the current time, is the estimated value of the model output, is the system noise, and the formula can obtain the relationship between the prediction value (model estimated value) of the battery terminal voltage at the current time and the actual measured value, i.e. the discretized electrical model output equation.

[0098] The online parameter identification process using the FFRLS algorithm is as follows:

[0099] ;

[0100] where represents the prediction value of the model terminal voltage at the current time, represents the transpose of the regression vector, represents the parameter estimation value at the previous time;

[0101] ;

[0102] where represents the terminal voltage prediction error;

[0103] ;

[0104] where represents the gain matrix at the current time, which is used to determine the weight of the prediction error on the parameter update, represents the parameter estimation covariance matrix at the previous time, is a forgetting factor, which controls the attenuation speed of the weight of old data, is a scalar, which is used to normalize the gain;

[0105] ;

[0106] where is the updated parameter estimation covariance matrix, represents the uncertainty reduction amount brought by the update;

[0107] ;

[0108] wherein, represents the updated parameter estimation vector, represents the parameter estimation vector at the last time, represents the parameter correction amount determined by the gain matrix and the prediction error.

[0109] The expression of the discretization of the lumped parameter thermal model is as follows:

[0110] ;

[0111] wherein and are the thermal resistance and the thermal capacity, , is the ambient temperature at the time, is the battery temperature at the time, is the heat generation at the time.

[0112] The regression vector and the parameter vector are expressed as follows:

[0113] ;

[0114] ;

[0115] wherein, the thermal identification model is as follows:

[0116] ;

[0117] wherein, is the model output temperature at the current time, is the estimation value of the model output, is the system noise, and the relationship between the prediction value (model estimation value) of the battery temperature at the current time and the actual measured value can be obtained from the formula, that is, the discretized thermal model output equation.

[0118] The online parameter identification process using the FFRLS algorithm is as follows:

[0119] ;

[0120] wherein is the temperature prediction value at the current time, represents the transpose of the regression vector, represents the estimated thermal model parameter vector at the last time;

[0121] ​​​ ;

[0122] wherein represents the current time temperature prediction error, represents the sensor measured temperature at the current time;

[0123] ;

[0124] wherein represents the current time gain matrix, used to determine the weight of the prediction error on the parameter update, represents the last time parameter estimation covariance matrix, is a forgetting factor, controlling the old data weight decay speed, is a scalar, used to normalize the gain; ;

[0125] wherein is the current time thermal model parameter vector, is the last time thermal model parameter estimation vector, is the parameter correction amount determined by the gain and the error.

[0126] S2, pulse charge-discharge experiments are performed on the lithium battery to obtain real-time terminal voltage and surface temperature data, and the collected terminal voltage and surface temperature data are input into the discretized electro-thermal-aging coupling model ;

[0127] S3, a double Kalman filter state observer is designed to observe the discretized electro-thermal-aging coupling model , and combined with the discretized electro-thermal-aging coupling model , the battery state of charge and the core temperature-aging joint state variable are estimated synchronously; the specific scheme is as follows:

[0128] Based on the unmeasurable state of charge value in the electrical model, a state observer is needed to estimate it in real time. The state equation and observation equation are as follows:

[0129] ;

[0130] ;

[0131] wherein, is the current time SOC prediction value, is the state transition function given by the equivalent circuit model, is the last time SOC posteriori estimation value, is the known input at the last time, is the actual measured value at the current time, The output function given for the model maps the SOC to the terminal voltage, The known input for the current time instant, is the state vector is the process noise and the measurement noise. This is the standard observation equation in the observer to produce the residual after a new measurement is obtained, which is compared to the predicted value to achieve the measurement update of the SOC.

[0132] The Unscented Kalman Filter is characterized by continuous-time propagation and discrete-time measurement update. The core procedure is as follows:

[0133] Generate Sigma points:

[0134] ;

[0135] From this step, we can get 2n+1 deterministic sampling points.

[0136] The expression of the continuous-time propagation based on each Sigma point is as follows:

[0137] ;

[0138] From this step, we can get the trajectory of each Sigma point in the continuous-time domain .

[0139] The propagated predicted state and covariance expressions are as follows:

[0140] ;

[0141] ;

[0142] where and are the weighting coefficients, is the process noise covariance matrix.

[0143] From this step, we can get the a priori state estimate and the a priori covariance .

[0144] The measurement update based on discrete time is as follows:

[0145] ;

[0146] From this step, we can get the output prediction corresponding to each Sigma point ;

[0147] ;

[0148] From this step, the predicted value can be observed ;

[0149] ;

[0150] From this step, the predicted covariance of observation can be obtained ;

[0151] ;

[0152] From this step, the cross covariance between state and observation can be obtained ;

[0153] ;

[0154] From this step, the Kalman gain matrix can be obtained ;

[0155] ;

[0156] Wherein is the observation variable, from this step, the updated state variable can be obtained ;

[0157] ;

[0158] From this step, the updated covariance can be obtained ;

[0159] Wherein is the predicted output, is the predicted observation mean, is the measurement noise covariance matrix, is the predicted observation covariance, is the cross covariance between state and observation, denotes the inverse of the matrix, denotes the transpose of the matrix, is the Kalman gain matrix, is the updated state variable, is the updated covariance.

[0160] S4, constant current constant voltage charging experiment is carried out on the lithium battery, four-dimensional target function is constructed taking charging current as control variable, considering charging time, capacity attenuation rate, current fluctuation rate and power deviation, and multi-constraint optimization problem is formed combining physical constraint and safety constraint of the battery;

[0161] By determining the boundary values of core temperature, capacity attenuation safety threshold and terminal voltage, the target function is constituted by input current, capacity attenuation, charging speed and dynamic power state, and the multi-objective and multi-constraint charging strategy based on model predictive control algorithm is constructed.

[0162] In the battery charging process, the maximum current, terminal voltage, capacity decay rate and core temperature of the battery need to be controlled below a reasonable threshold, and the charging constraints are set based on the above requirements:

[0163] ;

[0164] ;

[0165] ;

[0166] ;

[0167] wherein, is the maximum charging and discharging current, and are the upper and lower limits of the battery voltage, and are the upper and lower limits of the battery core temperature, is the maximum single-step capacity decay amount, is the nominal capacity.

[0168] The four-dimensional objective function expression is as follows:

[0169] ;

[0170] wherein, the symbol "estimation value, is the target SOC value, and respectively represent the target input current and the capacity loss term, and N is the prediction time domain, is the weighted square 2-norm, is the real-time charging power, is the power state upper limit P, Q, R, S at the current time, and the weight matrix needs to be adjusted. S5, the power state and the power deviation obtained by establishing the power state model and calculating are used to construct the weight matrix of the four-dimensional objective function according to the power state, the power deviation, and the state variables of the state of charge and the core temperature-aging joint state;

[0171] S6, a quadratic programming solver is used to solve the model multi-constraint optimization problem at the current time to generate an optimal charging sequence. The specific implementation is as follows: a quadratic programming solver is used to solve the model predictive optimization problem at the current time, based on the feedback of the state observer in the current closed-loop system, the weight matrix is dynamically adjusted according to the real-time state of charge, the power state and the deviation, and an optimal current sequence is generated.

[0172]

[0173] ​The objective function and linear constraint conversion standard quadratic programming form as follows:

[0174]

[0175] wherein, denotes the transpose matrix of is the optimization variable vector, the matrix is the Hessian matrix, denotes the transpose matrix of is the linear coefficient vector, is the equality constraint matrix, is the equality constraint vector, is the inequality constraint matrix, is the inequality constraint vector, is the variable upper and lower bound vector.

[0176] The quadratic programming (QP) form converts the real-time optimization problem of model predictive control (MPC) of the model prediction control (MPC) into a standard mathematical framework, that is, dynamically adjusts the weight matrix according to the state observer feedback of the state of charge, power and deviation, and comprehensively considers the terminal voltage, temperature, aging rate and power constraint; by solving the QP problem efficiently, the optimal current prediction sequence in the future time domain is generated, and only the first current value in the sequence is taken as the optimal charging current instruction actually applied at the current time, which dynamically tracks the target while meeting all safety constraints through closed-loop feedback rolling optimization, overcomes model errors and avoids voltage, temperature, aging or power overshoot risk.

[0177] The optimization solution of the fast charging decision current is input into the closed-loop feedback system based on model predictive control to avoid battery terminal voltage, core temperature, aging rate, power overshoot caused by model error, and determine the current time of electric vehicle fast charging current.

[0178] In this specific embodiment, the current of the model predictive control (MPC) charging strategy obtained by simulation is applied to the real battery charging process in this patent embodiment. As shown in Figure 5 , in summary, although the CCCV method is simple and reliable, it has limitations in dynamic adjustment and constraint processing, and it takes a long time to complete charging in the high state of charge region because the current in the constant voltage stage will gradually decrease, resulting in reduced charging efficiency. In contrast, the MPC strategy can provide a better balance between robustness and response speed in the battery system, effectively optimizing the changes in voltage, temperature, current, SOP and SOC, thereby improving charging efficiency and battery life.

[0179] ​It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and implementations, it is to be understood that the terminology used is for the purpose of descriptive clarity and that it is intended to be limited only by the words recited in the appended claims. The scope of the present application shall be limited only by the claims.

[0180] While the embodiments of the application have been shown and described herein, it is to be understood that the application is not limited to these embodiments. Rather, numerous modifications are possible without departing from the spirit and scope of the present application as delineated by the claims and their equivalents.

Claims

1. A dynamic power state-based model predictive control charging optimization method, comprising the following steps, characterized in that: S1, establish an electro-thermal-aging coupling model of a lithium battery , and discretize the electro-thermal-aging coupling model to obtain a discretized electro-thermal-aging coupling model , and discretize the electro-thermal-aging coupling model to obtain a discretized electro-thermal-aging coupling model ; S2, pulse charge-discharge experiment is carried out on the lithium battery to obtain real-time terminal voltage and surface temperature data, and the collected terminal voltage and surface temperature data are input into the discretized electro-thermal-aging coupling model ; S3, design double Kalman filter state observer for the discretized electro-thermal-aging coupling model Perform observation, and combine the discretized electro-thermal-aging coupling model Synchronize to estimate the battery state of charge and core temperature-aging joint state variables; S4, constant current and constant voltage charging experiment is carried out on the lithium battery, a four-dimensional objective function considering charging time, capacity decay rate, current fluctuation rate and power deviation is constructed with charging current as control variable, and a multi-constraint optimization problem is formed in combination with physical constraints and safety constraints of the battery; S5, a power state model is established and the calculated power state and power deviation are obtained, and a weight matrix of the four-dimensional objective function is constructed according to the power state, the power deviation and the state variables of the state of charge and the core temperature-aging joint state; The power state model in the S5 is a multi-constraint joint model considering voltage, temperature and aging, and its expression is as follows: ； wherein, is the upper voltage limit, is the open circuit voltage, is the core temperature upper limit, is the core temperature, is the thermal resistance, is the battery current, is the capacity fade threshold, is the capacity fade value, is the battery ohmic internal resistance, and is the aging factor; S6, a quadratic programming solver is used to solve the model multi-constraint optimization problem at the current time, and an optimal charging sequence is generated.

2. The model predictive control charging optimization method based on dynamic power states according to claim 1, wherein, The electro-thermal-aging coupling model of the lithium battery in the S1 includes an electrical model, and the electrical model adopts a first-order RC equivalent circuit model, and the expression of the circuit model is as follows: ； ； wherein, denotes the derivative of the state of charge with respect to time, is the state of charge of the battery, is the input current of the battery, denotes the derivative of the polarization voltage with respect to time, denotes the polarization voltage, denotes the rated capacity of the battery, and denote the polarization resistance and the polarization capacitance, respectively.

3. The model predictive control charge optimization method based on dynamic power states according to claim 1, wherein, The electro-thermal-aging coupling model of the lithium battery in the S1 also includes a lumped parameter thermal model, and the change rate of the battery temperature is determined by the relationship between heat generation and heat dissipation, and its expression is as follows: ； wherein, denotes the derivative of the temperature with respect to time, is a heat generation term of the battery, is the ambient temperature, is the battery temperature; is the thermal resistance, is the thermal capacity.

4. The model predictive control charge optimization method based on dynamic power states of claim 1, wherein, The electro-thermal-aging coupling model of the lithium battery in the S1 also includes a semi-empirical aging model, and its expression is as follows: ； ； wherein is the battery charge-discharge rate, is the capacity fade amount, and is is the pre-exponential factor of the rate function, is the activation energy, is the basic number of the aging rate, is the sensitivity index of the aging rate to the dynamic rate, is the time step is the instantaneous current, is the universal gas constant, is the absolute temperature of the battery, represents the rated capacity of the battery; The power state model in the S5 is a multi-constraint joint model considering voltage, temperature and aging, and its expression is as follows: ； wherein, is the upper voltage limit, is the open circuit voltage, is the core temperature upper limit, is the core temperature, is the thermal resistance, is the battery current, is the capacity fade threshold, is the capacity fade value, is the battery ohmic internal resistance, and is the aging factor.

5. The model predictive control charging optimization method based on dynamic power states as claimed in claim 2, wherein, The discretization processing of the electrical model is to identify the electrical model parameters and the thermal model parameters on line by using the recursive least squares method with forgetting factor, and the method is as follows: The expression obtained after the discretization of the circuit model is as follows: ； wherein, and are the terminal voltage and current at time k, respectively, is the sampling time, is the ohmic internal resistance of the battery, mainly composed of the intrinsic resistance of the electrolyte and the electrode; and collectively characterize the polarization effect, describing the dynamic behavior of the battery during charging and discharging.

6. The model predictive control charging optimization method based on dynamic power states according to claim 3, wherein, The expression obtained after the discretization of the lumped parameter thermal model is as follows: ； wherein and is the thermal resistance and the thermal capacity, , is the ambient temperature at the moment, is the battery temperature at the moment, is the heat generation at the moment.

7. The model predictive control charging optimization method based on dynamic power states according to claim 1, wherein, The state of charge of the electrical model part in the S3 is used as the state variable of the system, and the unscented Kalman filter algorithm is used to construct a state observer, and its state equation and observation equation are as follows: ； ； wherein, is the current time SOC prediction value, is the state transition function given by the equivalent circuit model, is the previous time SOC posterior estimate, is the known input at the previous time, is the actual measurement at the current time, is the output function given by the model mapping the SOC to the terminal voltage, is the known input at the current time, is the current time SOC prediction value and is the process noise and the measurement noise.

8. The model predictive control charging optimization method based on dynamic power states of claim 1, wherein, The physical constraints and safety constraints of the battery in the S4 are as follows: ； ； ； ； wherein, is the maximum charge and discharge current, and upper and lower limits of the battery voltage, and upper and lower limits of the battery core temperature, is the maximum capacity stepwise attenuation amount, is the nominal capacity.

9. The model predictive control charging optimization method based on dynamic power states according to claim 1, wherein, The expression of the four-dimensional objective function in the S4 is as follows: ； wherein the symbol represents an estimated value, is a target SOC value, and respectively represent a target input current and a capacity loss term, N is a prediction time domain, is a weighted square 2-norm, is a real-time charging power, is a power state upper limit at the current time, Pm, Qm, Rm, S are a weight matrix that needs to be adjusted.

Images