Retired battery rapid capacity estimation method based on electrochemical impedance spectroscopy and partial discharge data

Abstract

The invention discloses a method for quickly estimating the capacity of a decommissioned battery based on electrochemical impedance spectroscopy and partial discharge data, which comprises the following steps of: 1, carrying out discharge test on a recycled decommissioned battery, and extracting the discharge data of the battery; 2, establishing a constant-current charging time prediction model so as to predict the capacity; and 3, correcting the initial capacity prediction by using the capacity estimation based on the internal resistance so as to realize more accurate battery capacity estimation. According to the method, the capacity of the decommissioned battery can be quickly and accurately estimated, and a reliable basis is provided for recycling and recycling management of the battery.

Description

Technical Field

[0001] The present invention relates to the technical field of retired battery recycling and secondary utilization, and in particular to a battery capacity estimation method. Background Art

[0002] As the new energy vehicle industry has been developing for more than a decade, the capacity of a large number of electric vehicle power batteries will be less than 80% of the initial capacity and will no longer be suitable for vehicles. These retired batteries are in huge quantities. If they are not handled well, they will inevitably cause waste of resources and environmental damage. For this reason, the cascade utilization of retired batteries is a good solution.

[0003] Due to the different use conditions of electric vehicles, the performance parameters of retired batteries are uneven. In order to make better use of these batteries, it is necessary to perform capacity detection and internal resistance detection on retired batteries. Usually, when recycling retired batteries, they are fully charged and discharged to obtain the capacity of the battery, so that batteries with similar capacity can be grouped for use, but this full charge and discharge process will consume a lot of time and energy. If the capacity test can be performed using part of the stored power when the battery is recycled, it will greatly reduce time and energy consumption and improve the efficiency of recycling and secondary use of retired batteries. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned scheme and provide a method for quickly estimating the capacity of retired batteries based on electrochemical impedance spectroscopy and partial discharge data, so as to achieve fast and accurate estimation of the capacity of retired batteries, and solve the problems of long testing time and high energy consumption of the full charge and discharge method used in traditional retired battery capacity estimation methods, thereby providing a reliable basis for battery reuse and recycling management.

[0005] In order to achieve the above-mentioned purpose, the present invention can achieve the purpose through the following technical solutions:

[0006] The method for rapidly estimating the capacity of retired batteries based on electrochemical impedance spectroscopy and partial discharge data of the present invention is characterized in that it is carried out according to the following steps:

[0007] Step 1: Collect discharge data of retired batteries:

[0008] Step 1.1 Collect the initial voltage of N retired batteries , and take the minimum value of the initial voltage of all retired batteries as the target voltage ,in, represents the initial voltage of the nth retired battery;

[0009] Step 1.2 Discharge N retired batteries at a constant current to the minimum cut-off voltage , and then discharge at a constant voltage until the current is less than the cut-off current Then stop, so as to collect the complete voltage and current data of N retired batteries;

[0010] Step 1.3 is the voltage sampling interval, collecting the voltage of N retired batteries from To minimum cut-off voltage Time series of changes ;in, represents the time series of the nth retired battery, and ,in, represents the time required for the nth retired battery to reach the sth interval; S represents the total number of intervals;

[0011] Step 2: Collect the electrochemical impedance spectrum of the retired battery and identify the parameters of the retired battery:

[0012] Step 2.1 Measure the electrochemical impedance spectra of N retired batteries and use them to establish a battery equivalent circuit model;

[0013] Step 2.2: Use the equivalent circuit parameter identification algorithm to identify the parameters of the battery equivalent circuit model, thereby obtaining the parameter vector of all retired batteries. ,in, represents the parameter vector of the nth retired battery, and ;in, represents the ohmic internal resistance of the nth retired battery, represents the SEI internal resistance of the nth retired battery, represents the charge transfer internal resistance of the nth retired battery;

[0014] Step 3: Collect charging data of retired batteries:

[0015] Step 3.1 Constant current charging process:

[0016] Charge N retired batteries with constant current to the maximum cut-off voltage , record the voltage of N retired batteries to reach Time required ;in, Indicates that the nth retired battery is from Constant current charging to Time required;

[0017] Step 3.2 Constant voltage charging process:

[0018] Charge N retired batteries at a constant voltage until the current is less than the cut-off current Then stop and obtain the capacity labels of N retired batteries ;in, Indicates the capacity label of the nth retired battery;

[0019] Step 4: Train the retired battery time prediction model optimized by the fishing optimization algorithm to obtain the first predicted capacity of the retired battery based on time:

[0020] Step 4.1 As the input of the LSTM network, As the output of the LSTM network, the LSTM network is trained to obtain a trained retired battery time prediction model, and the root mean square error between the true value and the predicted value is calculated as the fitness function;

[0021] Step 4.2: According to the fitness function, the fishing optimization algorithm is used to optimize the hyperparameters of the trained retired battery time prediction model, so as to obtain the retired battery time prediction model with the optimal hyperparameters;

[0022] Step 4.3 Input the optimal hyperparameters into the retired battery time prediction model and process it to obtain N retired batteries from Constant current charging to Optimal forecast time required ;in, Indicates that the nth retired battery is from Constant current charging to The optimal forecast time required;

[0023] Step 4.4 Calculate the first predicted capacity based on time of the nth retired battery according to formula (5): , thus obtaining the first predicted capacity sequence of N retired batteries ;

[0024] (5)

[0025] In formula (5), Indicates the conversion efficiency from actual charging capacity to battery capacity; Indicates constant current charging current;

[0026] Step 5: Obtain a second predicted capacity of the retired battery based on internal resistance;

[0027] Step 5.1 Obtain the internal resistance of the nth retired battery according to formula (6): :

[0028] (6)

[0029] Step 5.2 Calculate the internal resistance according to formula (7) The second predicted capacity :

[0030] (7)

[0031] In formula (7), is the average value of the initial internal resistance of the retired battery, is the average value of the initial capacity of retired batteries;

[0032] Step 6: Compensate the first predicted capacity based on time with the second predicted capacity based on internal resistance, so as to obtain the optimal estimated capacity of the nth retired battery using formula (8): :

[0033] (8)

[0034] In formula (8), is the proportionality coefficient.

[0035] An electronic device of the present invention includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the method for quickly estimating the capacity of retired batteries, and the processor is configured to execute the program stored in the memory.

[0036] The present invention provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and the computer program executes the steps of the method for rapidly estimating the capacity of retired batteries when the computer program is executed by a processor.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention combines the characteristics of retired batteries that are generally transported at a low SOC for transportation safety, and only uses the battery's stored power for testing, without the need for a complete charge and discharge process. Compared to obtaining battery capacity by full charge and discharge, this design significantly reduces energy consumption, shortens test time, and helps improve test efficiency.

[0039] 2. The present invention uses the equal voltage time corresponding to the partial discharge data to predict the constant current charging time of the complete battery, so as to obtain the estimated capacity based on time; secondly, the estimated capacity based on internal resistance is used to compensate for the estimated capacity based on time, thereby achieving optimal capacity estimation and improving the accuracy of the estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a constant current discharge voltage curve diagram of the battery of the present invention;

[0041] Figure 2 This is a diagram of an equivalent circuit model of a battery of the present invention;

[0042] Figure 3 Flow chart of optimizing LSTM for FOA of the present invention;

[0043] Figure 4 The battery capacity estimation flow chart of the present invention;

[0044] Figure 5 This is a diagram of the test system of the present invention. DETAILED DESCRIPTION

[0045] In this embodiment, a data-driven method for fast capacity estimation of retired batteries based on electrochemical impedance spectroscopy and partial discharge data uses the time corresponding to the same voltage change interval of the retired battery to predict the battery constant current charging time, thereby obtaining a time-based capacity estimation; secondly, the initial capacity prediction is corrected using the internal resistance-based capacity estimation, thereby achieving a more accurate battery capacity estimation. Specifically, Figure 4 As shown, the method mainly includes the following steps:

[0046] Step 1: If Figure 5 As shown, collect retired battery discharge data:

[0047] Step 1.1 Since the initial voltage of each retired battery is different, the voltage changes of all retired batteries during the discharge process are different. Collect the initial voltage of N retired batteries ;in, Indicates the initial voltage of the nth retired battery, and takes the minimum value of the initial voltage of all retired batteries as the target voltage , thus ensuring that all batteries have the same voltage range. Figure 1 The voltage represented by the dashed line indicates .

[0048] Step 1.2 Discharge N retired batteries at a constant current to the minimum cut-off voltage , and then discharge at a constant voltage until the current is less than the cut-off current Then stop, so as to collect the complete voltage and current data of N retired batteries;

[0049] Step 1.3 The voltage of N retired batteries is collected from To minimum cut-off voltage Time series of changes ;in, Represents the time series of the nth retired battery ,in, Indicates that the nth retired battery reaches the sth The time required for the interval. Here S depends on For example, =0.1V, =2.5V, =3V, then n=5. The discharge time of the retired battery voltage from 3V to 2.9V, It is the time it takes for the retired battery voltage to discharge from 2.9V to 2.8V, and so on.

[0050] Step 2: Collect the electrochemical impedance spectrum of the retired battery and identify the parameters of the retired battery:

[0051] Step 2.1 Use an electrochemical workstation or other battery impedance spectrum acquisition device to measure the electrochemical impedance spectrum of all retired batteries in the frequency range of 10000Hz~0.01Hz using the current excitation method, and Figure 2 Establish a battery equivalent circuit model;

[0052] Step 2.2 Use the equivalent circuit parameter identification algorithm to identify the parameters of the battery equivalent circuit model, thereby obtaining the parameter vector of all retired batteries. ,in, represents the parameter vector of the nth battery, and ;in, represents the ohmic internal resistance of the nth battery, represents the SEI internal resistance of the nth battery, Represents the charge transfer internal resistance of the nth battery.

[0053] Step 3: Collect charging data of retired batteries and use the battery data to train the model.

[0054] Step 3.1 Constant current charging process:

[0055] Charge N retired batteries with constant current to the maximum cut-off voltage , and record the voltage of N retired batteries reaching Time required ;in, Indicates that the nth retired battery is from Constant current charging to Time required.

[0056] Step 3.2 Constant voltage charging process:

[0057] Charge N retired batteries at a constant voltage until the current is less than the cut-off current Then stop and use the ampere-hour integration method to obtain the capacity labels of N retired batteries. ;in, Indicates the capacity label of the nth retired battery.

[0058] Step 4: Train the retired battery time prediction model optimized by the fishing optimization algorithm to obtain the first predicted capacity of the retired battery based on time:

[0059] Step 4.1: Change the voltage of N retired batteries in step 1.4 from To cut-off voltage Time series of changes As the input of the LSTM network, the voltage of N retired batteries arrives at Time required As output, the trained retired battery time prediction model is obtained, and the root mean square error between the true value and the predicted value is calculated as the fitness function;

[0060] Step 4.2: According to the fitness function, the fishing optimization algorithm is used to optimize the hyperparameters of the trained retired battery time prediction model, so as to obtain the retired battery time prediction model with the optimal hyperparameters, such as Figure 3 As shown;

[0061] Step 4.2.1. Combine the hyperparameters of the retired battery time prediction model into a vector to represent the position of each “fish”, including the global search fish. and local search for fish ; Set global search for fish schools The population is F, and the local search for fish is The population size is J; the threshold value of the change range of the fitness value is set And the number of iterations k, and initialize k=1; Indicates the maximum number of iterations;

[0062] Step 4.2.2 Randomly initialize the kth generation local search fish school center position .

[0063] (1)

[0064] in, is a random vector uniformly distributed between [0,1], and are the lower and upper bounds of each hyperparameter, respectively.

[0065] Step 4.2.3 For the jth local search fish of the kth generation, at the center of the current fish school According to formula (2), the position of the jth local search fish in the kth generation is obtained: :

[0066] (2)

[0067] in: is a random vector with small amplitude, Represents the local search step size.

[0068] Step 4.2.4 To avoid local optimality, the global search fish group needs to perform a global search in the entire hyperparameter space. According to formula (3), the position of the fth local search fish in the kth generation is obtained: :

[0069] (3)

[0070] Step 4.2.5: Use the root mean square error as the fitness function to calculate the fitness of all fish in the kth generation, and select the maximum value from all fitness values. The position of this fish is taken as the optimal position of the kth generation. .

[0071] Step 4.2.6 The k+1 generation local search for the center position of the fish school Towards the optimal position Move, and its position can be obtained by formula (4):

[0072] (4)

[0073] in, is the learning factor that controls the moving step size.

[0074] Step 4.2.7 After assigning k+1 to k, determine whether k is greater than the preset termination generation or the change in fitness value is lower than the threshold A. If the requirement is met, output the optimal position of the kth generation population. And as the model parameters, thus obtaining the retired battery time prediction model with the optimal hyperparameters; otherwise, go to step 4.2.3 to continue searching for the optimal solution.

[0075] Step 4.3 Input the optimal hyperparameters into the retired battery time prediction model and process it to obtain N retired batteries from Constant current charging to Optimal forecast time required ;in, Indicates that the nth retired battery is from Constant current charging to The optimal forecast time required;

[0076] Step 4.4 Calculate the first predicted capacity based on time of the nth retired battery according to formula (5): , thus obtaining the first predicted capacity sequence of N retired batteries ;

[0077] Step 4.4.1 The constant current charging capacity can be approximately calculated by multiplying the constant current charging current and time, but this capacity will be lost due to the internal impedance of the battery, heat generation and other electrochemical reactions, so it is necessary to correct the capacity. Calculate the corrected capacity according to formula (5): :

[0078] (5)

[0079] in, Indicates the conversion efficiency of charging energy to battery capacity, which can be determined based on experimental data or parameters provided by the battery manufacturer; Indicates constant current charging current.

[0080] Step 5: Obtain a second predicted capacity based on internal resistance estimation;

[0081] Step 5.1 The internal resistance parameter of the nth retired battery can be obtained through step 2.2 , the internal resistance of the nth retired battery can be obtained through formula (6).

[0082] (6)

[0083] Step 5.2 Calculate the internal resistance according to formula (7) The second predicted capacity ;

[0084] (7)

[0085] In formula (7), is the average value of the initial internal resistance of the retired battery, is the average value of the initial capacity of retired batteries. and It can be determined based on the parameters provided by the battery manufacturer;

[0086] Step 6: After constant current charging, the battery needs to be charged at constant voltage. During the constant voltage charging process, the current will gradually decrease. At this time, the estimated capacity based on internal resistance can be used to compensate the estimated capacity based on time. Thus, the optimal estimated capacity of the nth retired battery can be obtained using formula (8): :

[0087] (8)

[0088] In formula (8), is the scaling factor, which adjusts the compensation of the time capacity estimate based on the resistance estimated capacity.

[0089] In this embodiment, an electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0090] In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium, and the computer program executes the steps of the above method when executed by a processor.

Claims

1. A method for rapid capacity estimation of retired batteries based on electrochemical impedance spectroscopy and partial discharge data, characterized in that: The steps are as follows: Step 1: Collect discharge data of retired batteries: Step 1.1 Collect the initial voltage of N retired batteries , and take the minimum value of the initial voltage of all retired batteries as the target voltage ,in, represents the initial voltage of the nth retired battery; Step 1.2 Discharge N retired batteries at a constant current to the minimum cut-off voltage , and then discharge at a constant voltage until the current is less than the cut-off current Then stop, so as to collect the complete voltage and current data of N retired batteries; Step 1.3 is the voltage sampling interval, collecting the voltage of N retired batteries from To minimum cut-off voltage Time series of changes ;in, represents the time series of the nth retired battery, and ,in, represents the time required for the nth retired battery to reach the sth interval; S represents the total number of intervals; Step 2: Collect the electrochemical impedance spectrum of the retired battery and identify the parameters of the retired battery: Step 2.1 Measure the electrochemical impedance spectra of N retired batteries and use them to establish a battery equivalent circuit model; Step 2.2: Use the equivalent circuit parameter identification algorithm to identify the parameters of the battery equivalent circuit model, thereby obtaining the parameter vector of all retired batteries. ,in, represents the parameter vector of the nth retired battery, and ;in, represents the ohmic internal resistance of the nth retired battery, represents the SEI internal resistance of the nth retired battery, represents the charge transfer internal resistance of the nth retired battery; Step 3: Collect charging data of retired batteries: Step 3.1 Constant current charging process: Charge N retired batteries with constant current to the maximum cut-off voltage , record the voltage of N retired batteries to reach Time required ;in, Indicates that the nth retired battery is from Constant current charging to Time required; Step 3.2 Constant voltage charging process: Charge N retired batteries at a constant voltage until the current is less than the cut-off current Then stop and obtain the capacity labels of N retired batteries ;in, Indicates the capacity label of the nth retired battery; Step 4: Train the retired battery time prediction model optimized by the fishing optimization algorithm to obtain the first predicted capacity of the retired battery based on time: Step 4.1 As the input of the LSTM network, As the output of the LSTM network, the LSTM network is trained to obtain a trained retired battery time prediction model, and the root mean square error between the true value and the predicted value is calculated as the fitness function; Step 4.2: According to the fitness function, the fishing optimization algorithm is used to optimize the hyperparameters of the trained retired battery time prediction model, so as to obtain the retired battery time prediction model with the optimal hyperparameters; Step 4.3 Input the optimal hyperparameters into the retired battery time prediction model and process it to obtain N retired batteries from Constant current charging to Optimal forecast time required ;in, Indicates that the nth retired battery is from Constant current charging to The optimal forecast time required; Step 4.4 Calculate the first predicted capacity based on time of the nth retired battery according to formula (5): , thus obtaining the first predicted capacity sequence of N retired batteries ; (5) In formula (5), Indicates the conversion efficiency from actual charging capacity to battery capacity; Indicates constant current charging current; Step 5: Obtain a second predicted capacity of the retired battery based on internal resistance; Step 5.1 Obtain the internal resistance of the nth retired battery according to formula (6): : (6) Step 5.2 Calculate the internal resistance according to formula (7) The second predicted capacity : (7) In formula (7), is the average value of the initial internal resistance of the retired battery, is the average value of the initial capacity of retired batteries; Step 6: Compensate the first predicted capacity based on time with the second predicted capacity based on internal resistance, so as to obtain the optimal estimated capacity of the nth retired battery using formula (8): : (8) In formula (8), is the proportionality coefficient.

2. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the method for quickly estimating the capacity of retired batteries as claimed in claim 1, and the processor is configured to execute the program stored in the memory.

3. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for quickly estimating the capacity of retired batteries according to claim 1 are executed.

Images