CN107171035B - Lithium-ion battery charging method - Google Patents

Abstract

A kind of charging method the invention discloses lithium ion battery includes: the equivalent-circuit model by neural network lithium ion battery；Coupled Heat Transfer model is on the basis of equivalent-circuit model to establish initial electric-thermal coupling model；Multiple groups constant-current charge scheme is designed, mesuring battary is tested according to constant-current charge scheme, obtains multiple groups test data；Multiple groups multistage charging scheme is designed, the multiple groups multistage charging scheme for meeting the constraint condition of Model for Multi-Objective Optimization is sequentially input to target electric-thermal coupling model, result data is obtained；And iteration optimization is executed to result data based on Model for Multi-Objective Optimization, one group of Pareto optimal solution is chosen from multiple groups multistage charging scheme as charging scheme to be selected；By decision-making technique from the charging scheme to be selected selection target charging scheme.The present invention can provide optimal charging scheme using limited experiment number for battery charging.

Description

Lithium-ion battery charging method

technical field

The invention relates to the technical field of batteries, in particular to a charging method for lithium ion batteries.

Background technique

Lithium-ion batteries have the advantages of high battery voltage, large specific capacity, low self-discharge effect, long cycle life, and no memory effect in secondary batteries. They are the best secondary batteries used in portable electronic devices and electric vehicles. type. However, lithium-ion batteries also have shortcomings such as easy overcharging and obvious aging, and there are certain safety issues when using them. In the process of work, it is necessary to scientifically and effectively control battery parameters and design a suitable charging strategy to ensure safe and efficient operation of the battery.

The charging strategy of the battery is a comprehensive issue. In order to increase the charging efficiency, the charging current must be increased; a large charging current will lead to an increase in the internal polarization of the battery and accelerate the aging rate of the battery cycle. Therefore, charging time, charging efficiency, and cycle life can be regarded as a set of trade-offs in the charging process. For charging strategies, it is difficult to design to improve the above charging performance in all aspects.

The most commonly used secondary battery charging method is constant current and constant voltage charging. However, during the constant current charging process, the battery's ability to withstand large currents gradually decreases, and the thermal effect gradually increases. Simply performing one-stage charging has defects in battery safety and energy efficiency. Therefore, the concept of multi-stage constant current charging is rejected. propose. This method is to divide the charging process into multiple stages, and try to make the constant current charging current in each stage adapt to the maximum acceptable charging current in the current battery. Therefore, the current in multiple stages is gradually reduced to achieve the purpose of ensuring the charging rate while protecting the safety of the battery. Studies have shown that the strategy of multi-stage constant current charging has the advantage of prolonging the battery cycle life compared with traditional constant current and constant voltage charging. The implementation of the multi-stage constant current charging method requires the design of the number of stages, the magnitude of the stage current, and the duration of the entire process. According to research, the current design of the multi-stage constant current strategy is mostly realized through experience. However, this empirical numerical design does not meet the application status quo of a wide variety of batteries and uneven performance indicators, and cannot meet the requirements of energy saving, consumption reduction, and charging rate improvement.

Contents of the invention

The technical problem to be solved by the present invention is to overcome that the charging method of lithium-ion batteries in the prior art either adopts constant current charging, or designs the value of multi-stage constant current charging through experience so that it cannot meet the requirements of energy saving, consumption reduction, and charging rate improvement. The invention provides a method for charging a lithium-ion battery.

The present invention solves the above technical problems through the following technical solutions:

A charging method for a lithium-ion battery, characterized in that the charging method may further comprise the steps:

S _1. Establishing an equivalent circuit model of a lithium-ion battery through a neural network; the input parameters of the neural network include battery temperature and battery state of charge data at different battery temperatures, and the output parameters include component parameters in the equivalent circuit model;

S _2. On the basis of the equivalent circuit model, the heat transfer model coupled with the natural convection of the battery is established to establish an initial electric-thermal coupling model;

S _3. Design multiple sets of constant current charging schemes, test the battery to be tested according to the constant current charging schemes, and obtain multiple sets of test data; the test data are used to fit the initial electric-thermal coupling model to obtain the target electric-thermal coupling model; Fitting parameters include input weights in the neural network and heat capacity and convective heat transfer coefficients in the heat transfer model;

Each set of constant current charging schemes includes charging time and corresponding current value; each set of test data includes terminal voltage, charging current, battery temperature and ambient temperature during the battery test process;

S _4. Design multiple sets of multi-stage charging schemes, each set of multi-stage charging schemes includes the number of charging stages, the charging time of each charging stage and the corresponding current value;

Input the multi-stage charging schemes that meet the constraints of the multi-objective optimization model into the target electric-thermal coupling model in turn to obtain the result data corresponding to each multi-stage charging scheme; and perform iterative optimization on the result data based on the multi-objective optimization model, Select a group of Pareto optimal solutions from the multiple groups of multi-stage charging schemes as the charging scheme to be selected;

Each set of result data includes the total charging time, the battery temperature difference before and after charging, and the energy loss rate during the charging process; the evaluation objectives of the multi-objective optimization model include the total charging time, the battery temperature difference before and after charging, and the energy loss rate during the charging process ;

S ₅ . Select a target charging scheme from the candidate charging schemes by using a decision-making method.

Preferably, the equivalent circuit model is a first-order RC branch equivalent circuit model, and elements in the first-order RC branch equivalent circuit model include DC resistance, polarization resistance and polarization capacitance;

The polarization resistance and the polarization capacitance form a parallel branch and are connected in series with the direct current resistance.

Preferably, in step S1, the element parameters include _a DC resistance value, a polarization resistance value and a capacitance value of a polarization capacitor.

Preferably, in step S1, the neural network is _an ELM neural network.

Preferably, in step S3, the initial electric _- thermal coupling model is fitted based on the Baron (Branch-And-Reduce Optimization Navigator, branch-and-reduce optimization navigation) algorithm.

Preferably, in step S5, the constraints of the multi _- objective optimization model include:

The current value of each charging stage decreases gradually;

Charge the battery to be tested until the SOC (the fraction of the charge capacity in the current battery to the maximum battery capacity) is greater than the first preset value;

During the charging process, the terminal voltage of the battery to be tested is less than or equal to the cut-off voltage;

During the charging process, the battery temperature is less than or equal to the second preset value.

Preferably, before step _S5 , the charging method further includes:

A multi-objective optimization model is established based on the multi-objective genetic algorithm.

Preferably, step _S6 specifically includes:

Select the target charging scheme from the charging schemes to be selected by a decision-making method based on the TOPSIS (a medium quality evaluation algorithm) algorithm and the entropy coefficient of the battery to be tested;

The entropy coefficient is the ratio of the terminal voltage of the battery to be tested to the battery temperature.

The positive and progressive effect of the present invention is that the present invention can provide the best charging scheme for battery charging with a limited number of experiments, so as to increase the charging rate, reduce the energy consumption of the charging process, and control the temperature increase of the battery.

Description of drawings

FIG. 1 is a flowchart of a charging method for a lithium-ion battery according to a preferred embodiment of the present invention.

Fig. 2 is a circuit diagram of the electric-thermal coupling model in Fig. 1 .

Detailed ways

The present invention is further illustrated below by means of examples, but the present invention is not limited to the scope of the examples.

As shown in Figure 1, the charging method of the lithium ion battery of the present embodiment comprises the following steps:

Step 101, establishing an equivalent circuit model of a lithium-ion battery through a neural network.

Among them, the input parameters of the neural network include the battery temperature and the state of charge of the battery at different battery temperatures, and the output parameters include the component parameters in the equivalent circuit model. In this embodiment, the ELM neural network is used for modeling. Specifically, as shown in Figure 2, the equivalent circuit model is a first-order RC branch equivalent circuit model, and the first-order RC branch equivalent circuit model includes DC resistance, polarization resistance and polarization capacitance, and the polarization resistance and The polarized capacitance forms a parallel branch and is connected in series with the DC resistance. The component parameters include the resistance value of the DC resistance, the resistance value of the polarization resistance and the capacitance value of the polarization capacitor.

Step 102 , on the basis of the equivalent circuit model, couple the battery natural convection heat transfer model to establish an initial electric-thermal coupling model.

In order to improve the accuracy of the mathematical model, the influence of temperature and SOC on the circuit parameters is considered in the modeling, so the equivalent circuit model and the thermal model are also coupled.

Step 103 , designing multiple sets of constant current charging schemes, testing the battery to be tested according to the constant current charging schemes, and obtaining multiple sets of test data; the test data are used to fit the initial electric-thermal coupling model to obtain the target electric-thermal coupling model.

Among them, the fitting parameters include the input weights in the neural network and the heat capacity and convective heat transfer coefficient in the heat transfer model; each set of constant current charging schemes includes charging time and corresponding current values; each set of test data includes the battery test The terminal voltage, charging current, battery temperature and ambient temperature during the process. In this embodiment, the initial electric-thermal coupling model is fitted based on the Baron algorithm.

The following is a detailed introduction to the process of mathematical modeling:

Before establishing the mathematical model, firstly, investigate and obtain the rated operating parameters of the target rechargeable battery, mainly including the maximum charging current, charging cut-off voltage, etc.; secondly, determine the possible operating temperature range of the battery; finally, design the battery operating temperature range according to the maximum charging current Constant current and constant voltage charging experiment, the more experimental data, the more accurate the mathematical model. In this embodiment, design 3-5 temperatures, 3-5 current values in the constant current stage, and carry out the CC-CV charging experiment starting at the selected temperature. Record the battery’s open circuit voltage, terminal voltage, charging current, real-time temperature, battery charge state at different temperatures and other data at a certain time interval, so that the U _OC -SOC curve and the entropy coefficient -SOC curve can be obtained. In order to further improve the accuracy of the model degree, and also perform a linear fit to the curve. Among them, SOC is the state of charge, between 0 and 1, and is also commonly expressed as a percentage, which refers to the fraction of the current charge capacity in the battery to the maximum battery capacity; U _OC is the open circuit voltage; the entropy coefficient is the terminal voltage of the battery and Ratio of battery temperature.

After obtaining the above sample training parameters, perform modeling.

Step 1: Establish a first-order RC equivalent circuit model to describe the dynamic characteristics of the battery. This model is composed of R1, C connected in parallel, R0 in series and U _OC of the battery;

The model can be described by the following equation:

U _b =U _OC +U ₀ +U ₁

Among them, U _b is the battery terminal voltage, U _OC is the open circuit voltage of the battery, U ₀ is the voltage at both ends of the DC resistance R ₀ , and U ₁ is the voltage at both ends of the RC parallel branch. This formula describes the terminal voltage and the open circuit voltage measured directly by the battery Voltage, voltage across model resistors.

The voltage value on the RC parallel branch is related to the size of the resistance and capacitance, which is derived according to Kirchhoff's current law and satisfies the following differential equation:

Among them, I is the total current on the main road, and the iterative formula can be obtained after integrating the above formula:

In the formula, τ=R ₁ *C ₁ is the time constant on the parallel circuit.

The thermal model part mainly considers the reversible heat and irreversible heat of the internal reaction of the battery. The heat generated by the battery is the energy of irreversible heat, that is, the energy loss can be expressed as:

The temperature rise of the battery comes from the difference between heat generation and convective heat transfer dissipation during charging, so the temperature satisfies the following differential equation:

Among them, S is the entropy coefficient, m is the mass of the battery, C _p is the heat capacity of the battery, h is the heat transfer coefficient of the battery surface, A is the heat transfer surface area, and Tem _amb is the ambient temperature; same as the voltage, the battery temperature can be obtained after integrating the differential equation Tem's iterative formula:

Wherein, Δt=t(n)-t(n-1).

Step 2: Obtain the input and output models of the system from the ELM neural network structure including the input layer, output layer and hidden layer.

The multi-output model can be expressed as follows:

where a _i and b _i are the input weights of the model, g(a _i , b _i , x) is the activation function of the ELM, β _ij is the output weight of a hidden node parameter, n is the number of hidden nodes in the model, combined The activation function of the following form, and the dual input variables of temperature and SOC, can be derived to express the general formula of the equivalent circuit parameters R ₀ , R ₁ , C, and h and C _p are taken as constants:

Due to the multi-input of the model, the input weight a _i includes two parts Iwi1 and Iwi2, and Ow is the output weight obtained after identification (fitting). In this method, the number of hidden nodes n is artificially designed, and Iwi1, Iwi2, and b _i are randomized by the algorithm. get.

Step 3: Calculate the SOC and open circuit voltage based on the test data in the battery test. The calculation of SOC uses the ampere-hour method, and the formula is expressed as:

C _i is the measured total capacity of the battery.

Step 4: The SOC, Temh, and Q calculated in steps 1, 2, and 3 and the above test parameters obtained through battery testing are used to fit and optimize the initial electric-thermal coupling model based on the Baron algorithm, and the least square method can be used. The optimization goal is to minimize the terminal voltage and temperature errors between the fitting value and the test value. The objective function can be expressed as:

in, refers to the fitted terminal voltage data, In the same way, the program calculates the optimal solution of the output weight of a set of ELM models, which is used to characterize the above-mentioned electric-thermal coupling model, so as to obtain the target electric-thermal coupling model.

Step 104, designing multiple sets of multi-stage charging schemes, inputting the multi-stage charging schemes satisfying the constraints of the multi-objective optimization model into the target electric-thermal coupling model in sequence, and obtaining the result data corresponding to each multi-stage charging scheme; and based on The multi-objective optimization model performs iterative optimization on the result data, and selects a set of Pareto optimal solutions from multiple sets of multi-stage charging schemes as the charging scheme to be selected.

Among them, each group of multi-stage charging schemes includes the number of charging stages, the charging time of each charging stage and the corresponding current value; each group of result data includes the total charging time, the battery temperature difference before and after charging, and the energy loss rate during the charging process; The evaluation objectives of the objective optimization model include the total charging time, the battery temperature difference before and after charging, and the energy loss rate during charging.

In this embodiment, a multi-objective optimization model is established based on a multi-objective genetic algorithm. The multi-objective optimization model performs iterative optimization on the result data, so that a group of Pareto optimal solutions that meet the evaluation objectives are selected from multiple sets of multi-stage charging schemes as the charging scheme to be selected.

Specifically, the multi-objective function of the multi-objective optimization model is expressed as:

minF ₂ =max(Tem)-Tem(0)

F ₁ , F ₂ , and F ₃ are the optimization goals during the charging process of the battery. F ₁ describes the sum of the charging time of all stages in a scheme, that is, the total charging time. F ₂ is the temperature difference of the battery before and after charging, and F3 is the charging time. The ratio of the sum of irreversible heat in the process to the total charging energy describes the energy loss rate.

The constraints that are met include:

I ₁ >I ₂ >… >I _k

U _b ≤ U _cutoff Tem ≤ Tem _max

Constraint 1 ensures that the current decreases step by step; Constraint 2 ensures that the solution strategy of the optimization calculation can charge the battery until the SOC is greater than the first preset value, and the first preset value is 0.99 as the best value, which is considered full; The terminal voltage cannot exceed the cut-off voltage (U _cutoff ), and the temperature cannot exceed the second preset value, that is, the maximum safety limit of the battery temperature (Tem _max ), which are the basic limits for battery safety considerations. The optimization variable is the number of stages of multi-stage constant current charging, the current of each stage, and the time of each stage. The goal of the algorithm is to optimize the overall performance of multi-stage constant current charging under the condition of satisfying the above constraints, and search for a suitable initial charging globally. The scheme is used as the charging scheme to be selected.

Step 105. Select a target charging scheme from the charging schemes to be selected by a decision-making method.

Specifically, this embodiment adopts a decision-making method based on the integration of entropy coefficient and TOPSIS to obtain the final multi-objective optimal charging strategy. In evaluating multi-objective synthesis problems, the decision-making steps are as follows:

Step 1: In the solution set of the optimization algorithm, filter out the optimal value of each optimization objective;

Wherein, i=1,2,...,p (P is the number of charging schemes to be selected); j=1,2,3.

The optimization goals are divided into two types: the bigger the better and the smaller the better. The goals in this problem are the smaller the better;

Step 2: Calculate the closeness of the target of each solution to the optimal value and perform normalization processing, and calculate the entropy weight of each index;

In the formula, d _ij is the normalization index of element y _ij , and E _j is the entropy value of the jth evaluation index;

Step 3: Combine the entropy weight with the subjective weight determined subjectively (can be set by yourself), and finally determine the weight of each indicator;

Objective weight:

Final weights:

where ω _j is the subjective weight of target j;

Step 4: Calculate the weighted Euclidean distance of the index. The shorter the distance, the better, and obtain the optimal solution under this decision, that is, the target charging scheme.

In this embodiment, a scheme for obtaining an optimal charging strategy from a battery is completely designed, and a charging strategy with better multi-objective performance is obtained. The model established by this method has sufficient accuracy, can well describe the dynamic characteristics of the process object, and has a fast calculation speed, which is expected to be improved to an online optimization mode.

Although the specific implementation of the present invention has been described above, those skilled in the art should understand that this is only an example, and the protection scope of the present invention is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principle and essence of the present invention, but these changes and modifications all fall within the protection scope of the present invention.

Claims (8)

1. a charging method of lithium ion battery, is characterized in that, described charging method comprises the following steps: S _1. Establishing an equivalent circuit model of a lithium-ion battery through a neural network; the input parameters of the neural network include battery temperature and battery state of charge at different battery temperatures, and the output parameters include component parameters in the equivalent circuit model; S _2. On the basis of the equivalent circuit model, couple the heat transfer model of battery natural convection to establish an initial electric-thermal coupling model; S _3. Design multiple sets of constant current charging schemes, test the battery to be tested according to the constant current charging schemes, and obtain multiple sets of test data; the multiple sets of test data are used to fit the initial electric-thermal coupling model to obtain the target electric-thermal coupling model; fitting parameters include input weights in the neural network and heat capacity and convective heat transfer coefficients in the heat transfer model; Each set of constant current charging schemes includes charging time and corresponding current value; each set of test data includes terminal voltage, charging current, battery temperature and ambient temperature during the battery test process; S _4. Design multiple sets of multi-stage charging schemes, each set of multi-stage charging schemes includes the number of charging stages, the charging time of each charging stage and the corresponding current value; Input the multi-stage charging schemes that meet the constraints of the multi-objective optimization model into the target electric-thermal coupling model in turn to obtain the result data corresponding to each multi-stage charging scheme; and perform iterative optimization on the result data based on the multi-objective optimization model, Select a group of Pareto optimal solutions from the multiple groups of multi-stage charging schemes as the charging scheme to be selected; Each set of result data includes the total charging time, the battery temperature difference before and after charging, and the energy loss rate during the charging process; the evaluation objectives of the multi-objective optimization model include the total charging time, the battery temperature difference before and after charging, and the energy loss rate during the charging process ; S ₅ . Select a target charging scheme from the candidate charging schemes by using a decision-making method. 2. the charging method of lithium-ion battery as claimed in claim 1 is characterized in that, described equivalent circuit model is a first-order RC branch equivalent circuit model, and in the described first-order RC branch equivalent circuit model Components include DC resistance, polarization resistance and polarization capacitance; The polarization resistance and the polarization capacitance form a parallel branch and are connected in series with the direct current resistance. 3. The charging method of lithium-ion battery as claimed in claim ₂ , characterized in that, in step S1, the component parameters include a DC resistance value, a polarization resistance value and a capacitance value of a polarization capacitor. 4. The charging method of lithium-ion battery as claimed in claim ₁ , characterized in that, in step S1, the neural network is an ELM neural network. 5. The charging method of lithium-ion battery as claimed in claim ₁ , characterized in that, in step S3, the initial electric-thermal coupling model is fitted based on the Baron algorithm. 6. The charging method of lithium-ion battery as claimed in claim ₁ , is characterized in that, in step S4, the constraint condition of multi-objective optimization model comprises: The current value of each charging stage decreases gradually; Charging the battery to be tested until the SOC is greater than the first preset value; During the charging process, the terminal voltage of the battery to be tested is less than or equal to the cut-off voltage; During the charging process, the battery temperature is less than or equal to the second preset value. 7. The charging method of lithium-ion battery as claimed in claim ₁ , is characterized in that, before step S4, described charging method also comprises: A multi-objective optimization model is established based on the multi-objective genetic algorithm. 8. The charging method of lithium ion battery as claimed in claim ₁ , is characterized in that, step S5 comprises: Select the target charging scheme from the charging scheme to be selected by a decision-making method based on the TOPSIS algorithm and the entropy coefficient of the lithium-ion battery; The entropy coefficient is the ratio of the terminal voltage of the battery to the battery temperature.