A method and system for delaying battery aging based on matched aging modes

Abstract

This invention provides a battery aging delay method and system based on matched aging modes, belonging to the field of battery aging delay technology. To address the problems of existing direct detection methods for battery aging, which require high-end equipment and limit application scenarios, and indirect detection methods, which have low accuracy and lack aging suppression methods, this invention utilizes cyclic aging tests to obtain battery aging parameters, identifies parameters based on a pseudo-two-dimensional model, then matches aging modes, trains a neural network model based on the matching relationship, and inputs real-time external battery data to obtain the actual battery aging mode and corresponding data. Finally, it uses the coupling of fuzzy PID control and battery control quantities to precisely control battery aging. This invention does not affect the normal use of the battery and provides corresponding aging suppression methods, allowing for personalized solutions to battery aging problems, possessing practical application value and broad market application prospects.

Description

Technical Field

[0001] This invention relates to the field of battery delayed aging technology, and more specifically, to a battery aging delay method and system based on matched aging modes. Background Technology

[0002] With the booming development of the new energy vehicle industry, the number of pure electric vehicles is gradually increasing. The concentrated charging time of pure electric vehicles puts enormous pressure on the power grid (peak-valley difference), and since electricity prices are adjusted in real time according to electricity consumption fluctuations, charging costs for car owners also increase. Vehicle-to-grid (V2G) technology has significant advantages in solving these problems, not only reducing grid pressure but also decreasing charging costs for users. However, V2G technology requires frequent charging and discharging of the battery; if this technology is not utilized properly, it can damage the battery's health and cause it to fail faster. Accurate analysis of battery aging and appropriate methods to delay battery aging are necessary for the widespread adoption of V2G technology and for improving battery life. Battery aging manifests in various ways, primarily including solid electrolyte interphase (SEI) growth, lithium dendrite growth, electrolyte decomposition, current collector corrosion, and active material loss.

[0003] Currently, battery aging diagnosis methods are mainly divided into direct measurement and indirect calculation methods. Post-disassembly analysis can intuitively observe the aging reaction inside the battery and provide a more accurate explanation of battery aging. Electrochemical impedance spectroscopy (EIS) measures the corresponding current (or potential) response generated by the system by applying a small-amplitude sinusoidal potential (or current) disturbance signal, thereby obtaining an impedance spectrum and analyzing the battery aging phenomenon. Curve-based analysis can be divided into three steps: (1) obtaining a clear incremental capacity / differential voltage (IC / DV) curve from the battery charging curve; (2) extracting feature points (FOI), including the position or height of the peak, the area under the IC peak, etc.; (3) using a lookup table method to diagnose battery aging in the current BMS (Battery Management System). There are also model-based analysis methods, which describe battery aging in detail by constructing an aging model and solving the model parameters. In addition, the development of data-driven and machine learning has provided new ideas for the construction of battery aging models. Most battery aging suppression methods are based on a large number of experiments to find the relationship between external battery data and battery aging, and control it when it reaches a certain threshold, thereby actively suppressing battery aging. Luo Guoqing et al. suppressed the impact of V2G technology on the aging of electric vehicle power batteries by constraining the battery charge-discharge rate. They also considered the influence of SOC (State of Charge) when constraining the charge-discharge rate, thus including the depth of charge and discharge within the control range. Jalkanen et al. analyzed the effect of temperature on battery aging through post-disassembly analysis, finding that during cycle aging at different temperatures, lithium dendrite precipitation increases with increasing temperature, thus affecting battery aging. Curve-based analysis methods are often combined with post-disassembly analysis to obtain more reliable results.

[0004] However, post-disassembly analysis involves irreversible and destructive disassembly of the battery, which is complex, costly, and renders the battery unusable. While electrochemical impedance spectroscopy (EIS) avoids this destructive process, its reliance on complex equipment limits its application. Curve-based analysis requires limiting the current rate to obtain the IC / DV curve, and its practical application is hampered by temperature variations and high instrument precision requirements. In summary, direct measurement methods limit their application due to the demanding testing equipment. Existing indirect measurement methods, such as neural network models based entirely on external battery data, can accurately predict battery SOH (State of Health) values, but their lack of identification of internal aging patterns leads to a deviation from the actual SOH aging trajectory in the later stages.

[0005] In addition, existing technologies for suppressing battery aging mainly affect battery aging by changing the operating temperature, charge / discharge rate, and depth of charge / discharge, but they are still somewhat lacking in terms of control methods. Summary of the Invention

[0006] The technical problem to be solved by this invention is:

[0007] To address the issues that existing direct detection methods for battery aging have high equipment requirements and limited application scenarios, while indirect detection methods have low accuracy and lack aging suppression methods.

[0008] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0009] This invention provides a battery aging delay method based on matched aging modes, comprising the following steps:

[0010] S100. Conduct a cycle aging test on the battery to obtain the battery aging condition data. Use the condition data as the result of battery cycle aging. Based on the pseudo two-dimensional model, perform parameter identification to obtain the internal state of the lithium-ion battery.

[0011] S200: Match the aging mode using the parameter identification results obtained in step S100. The aging mode includes SEI film growth, lithium dendrite formation, electrolyte decomposition, and active material loss.

[0012] S300: The matching analysis results obtained in step S200 are used as training data and input into the radial basis neural network for training to obtain the trained neural network model.

[0013] S400: Using the neural network model trained in step S300, input real-time power battery detection data to determine the state of the battery, and then analyze the aging phenomena that will occur in the battery in the future, and carry out corresponding control for different aging phenomena.

[0014] Furthermore, in step S100, when performing parameter identification using the pseudo-two-dimensional model, the following steps are included:

[0015] Charge conservation equation for solid particles:

[0016]

[0017] Where, φ s Represents the potential of a solid particle; σ eff Represents the effective solid particle conductivity; F represents the Faraday constant; j represents the ion flux; a s denoted by , where represents the surface area of ​​the solid particle; x represents the spatial coordinates of the internal potential change of the solid particle.

[0018] The mass conservation equation for solid particles:

[0019]

[0020] Among them, c s Represents the lithium concentration in the solid phase (related to position r and time t); t represents time; r represents the radial coordinate of the particle;

[0021] Represents the effective solid particle diffusion rate;

[0022] Electrolyte charge conservation equation:

[0023]

[0024] Among them, κ eff Represents the effective electrolyte conductivity; φ e Represents electrolyte potential; R represents universal gas constant; T represents temperature; t + Represents the migration number; c e This represents the lithium concentration in the electrolyte phase;

[0025] Electrolyte mass conservation equation:

[0026]

[0027] Where, ε p,n Represents porosity; Represents the effective electrolyte diffusivity; ε s The volume percentage of solid particles;

[0028] The equation of lithium-ion motion between solid particles and electrolyte:

[0029]

[0030] Where, k p,n Represents the reaction rate constant; This represents the maximum concentration of lithium in the solid phase. η represents the lithium concentration at the solid surface; η represents the overpotential; and α represents the transfer coefficient.

[0031] Further, in step S200,

[0032] For SEI film growth: effective electrolyte conductivity κ eff and effective electrolyte diffusion rate Changes have occurred;

[0033] For lithium dendrite formation: effective electrolyte conductivity κ eff Effective electrolyte diffusion rate solid particle surface area a sEffective solid particle diffusion rate And effective solid particle conductivity σ eff Changes have occurred;

[0034] For electrolyte decomposition: effective electrolyte diffusivity Changes have occurred;

[0035] For active material loss: effective solid particle diffusion rate With the surface area a of solid particles s Things have changed.

[0036] Furthermore, in step S400, the real-time acquired detection data includes battery temperature, depth of charge / discharge, and charge / discharge rate. Based on fuzzy PID control and considering the coupling of battery control quantities, the control quantities are decoupled before control is implemented to control battery aging.

[0037] Furthermore, in step S400, when conducting a cycle aging experiment on the battery, the corresponding external signal data of the battery is obtained through cycle aging tests at different temperatures, and the data is input into the battery pseudo-two-dimensional model for parameter identification; the fitness function is the minimum root mean square of the model voltage and the measured voltage, and a genetic algorithm is used for parameter identification.

[0038] A battery aging delay system based on a matched aging mode is provided. The system has a program module corresponding to the steps described above, and executes the steps in the battery aging delay method based on a matched aging mode described above when running.

[0039] A computer-readable storage medium storing a computer program configured to, when invoked by a processor, implement steps of a battery aging delay method based on a matched aging pattern.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] This invention discloses a battery aging delay method and system based on matched aging modes. It utilizes cyclic aging tests to obtain battery aging parameters, identifies these parameters using a pseudo-two-dimensional model, matches them to aging modes, trains a neural network model based on the matching relationships, and inputs real-time external battery data to obtain the actual battery aging mode and corresponding data. Finally, it employs the coupling of fuzzy PID control and battery control quantities to precisely control battery aging. This invention eliminates the need for battery disassembly, analyzes battery aging based on external data monitoring, has low requirements for testing equipment, does not affect normal battery use, and provides relatively accurate analysis results. It can also yield corresponding aging suppression methods, and supplementary suppression methods can be used to address battery aging problems in a personalized manner, demonstrating practical application value and broad market application prospects. Attached Figure Description

[0042] Figure 1 This is a flowchart of a battery aging delay method based on a matched aging mode, as described in an embodiment of the present invention. Detailed Implementation

[0043] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0044] Specific Implementation Plan 1: Combining Figure 1 As shown, the present invention provides a battery aging delay method based on a matched aging mode, comprising the following steps:

[0045] S100. Conduct cycle aging tests on ternary lithium batteries to obtain battery aging condition data. Use the condition data as the result of battery cycle aging. Based on a pseudo-two-dimensional model, perform parameter identification to obtain the internal state of the lithium-ion battery.

[0046] During the battery cycle aging test, the corresponding external signal data of the battery is obtained through the battery cycle aging test, input into the battery pseudo two-dimensional model and perform parameter identification; the fitness function is the minimum root mean square of the model voltage and the measured voltage, and the parameter identification is performed using a genetic algorithm.

[0047] The aging cycle steps are as follows:

[0048] (1) Set the constant temperature chamber to 25℃ and place the battery in the constant temperature chamber until the temperature stabilizes;

[0049] (2) Charge the battery using CCCV (constant current and constant voltage) method. Charge the battery with a constant current of 1C until the upper cutoff voltage is reached, then switch to constant voltage charging until the cutoff current is 0.05C. Stop charging and let it rest for 10 minutes after stopping charging.

[0050] (3) Discharge at a constant current rate of 3C to the lower cutoff voltage;

[0051] (4) Repeat steps (2)-(3) 10 times;

[0052] (5) Set the constant temperature chamber to 25°C, put the battery into the constant temperature chamber until the temperature stabilizes, recharge the battery with CCCV and then perform a discharge capacity test, and record the capacity under the current cycle.

[0053] (6) Repeat steps (1)-(5) until the measured discharge capacity is less than 80% of the initial capacity;

[0054] (7) Replace the battery cell and repeat steps (1)-(6);

[0055] Note: At this time, the discharge rate current is determined based on the actual maximum capacity value obtained from the discharge capacity test.

[0056] When using a pseudo-two-dimensional model for parameter identification, it includes:

[0057] Charge conservation equation for solid particles:

[0058]

[0059] Where, φ s Represents the potential of a solid particle; σ eff Represents the effective solid particle conductivity; F represents the Faraday constant; j represents the ion flux; a s denoted by , where represents the surface area of ​​the solid particle; x represents the spatial coordinates of the internal potential change of the solid particle.

[0060] The mass conservation equation for solid particles:

[0061]

[0062] Among them, c s Represents the lithium concentration in the solid phase (related to position r and time t); t represents time; r represents the radial coordinate of the particle; Represents the effective solid particle diffusion rate;

[0063] Electrolyte charge conservation equation:

[0064]

[0065] Among them, κ eff Represents the effective electrolyte conductivity; φ e Represents electrolyte potential; R represents universal gas constant; T represents temperature; t + Represents the migration number; c e This represents the lithium concentration in the electrolyte phase;

[0066] Electrolyte mass conservation equation:

[0067]

[0068] Where, ε p,n Represents porosity; Represents the effective electrolyte diffusivity; ε s The volume percentage of solid particles;

[0069] The equation of lithium-ion motion between solid particles and electrolyte:

[0070]

[0071] Where, k p,n Represents the reaction rate constant; This represents the maximum concentration of lithium in the solid phase. Represents the lithium concentration at the solid surface; η represents the overpotential; α represents the transfer coefficient;

[0072] Parameter identification is performed using a genetic algorithm;

[0073] The parameters identified by the genetic algorithm are anode solid-phase diffusivity, anode solid-phase conductivity, cathode solid-phase diffusivity, and cathode solid-phase conductivity. The fitness function is the mean square value of the measured voltage and the estimated voltage. The genetic algorithm flowchart is as follows:

[0074] (1) Initialize the population: Use binary numbers to represent the range of each parameter to obtain the individual encoding of the population;

[0075] (2) Calculate individual fitness: After encoding and decoding the individuals in the population, calculate their fitness.

[0076] (3) Selection: Select individuals based on the fitness calculated in the previous step and retain them;

[0077] (4) Crossover: New individuals are generated through the crossover and fusion of individual codes;

[0078] (5) Mutation: New individuals are generated through low-probability mutations;

[0079] (6) Individual recombination generates the next generation of population: The new individuals obtained through steps (3)-(5) are combined together to form a new population;

[0080] (7) Iterate until the termination condition is met: Repeat steps (2)-(6) until the number of iterations reaches the requirement;

[0081] Using the above method to output parameter identification results, the parameter identification results under different SOH conditions of the battery are linked with SOH to find the range of identification parameters when aging occurs;

[0082] S200: Using the parameter identification results obtained in step S100, the aging mode is matched. This invention primarily uses four aging modes: SEI film growth, lithium dendrite formation, electrolyte decomposition, and active material loss.

[0083] For SEI film growth: The resistance and thickness of the SEI film affect the conductivity of the electrolyte and the diffusion of lithium ions. Therefore, the effective electrolyte conductivity κ in the above equation is... eff and effective electrolyte diffusion rate Things will change;

[0084] Regarding lithium dendrite formation: The formation and growth of lithium dendrites consume active lithium and electrolyte in the battery, leading to battery capacity degradation, increased internal resistance, and electrolyte depletion. Lithium dendrites expose more fresh lithium metal to the organic electrolyte, continuously generating new SEI and dendrites, which affects the effective electrolyte conductivity κ. eff Effective electrolyte diffusion rate solid particle surface area a s Effective solid particle diffusivity and effective solid particle electrical conductivity;

[0085] Regarding electrolyte decomposition: Electrolyte decomposition alters the composition and concentration of the electrolyte, thereby affecting the effective electrolyte diffusivity in the electrolyte mass conservation equation.

[0086] Regarding active material loss: A reduction in active material decreases the active surface area of ​​the electrode and the diffusion capacity of lithium ions; therefore, the effective solid particle diffusion rate in the model... With the surface area a of solid particles s Things will change;

[0087] To reduce the complexity of the calculations, only four charging parameters that are relatively close to battery capacity decay and voltage are selected for identification: anode solid phase diffusivity, anode solid phase conductivity, cathode solid phase diffusivity, and cathode solid phase conductivity.

[0088] To maximize the battery's usability, certain limitations should be added to the definition of each aging phenomenon. Only when the changes in each parameter exceed the design range should aging be considered to have occurred.

[0089]

[0090] S300: Using the matching analysis results obtained in step S200 as training data, input the data into the radial basis neural network for training to obtain the final neural network model. In subsequent aging analysis, parameter identification is not required frequently, reducing computation time and computational load, and simplifying the analysis process.

[0091] Using voltage and current features from the data as input and parameter changes as output, a radial basis function neural network is constructed.

[0092] Radial basis function neural network design:

[0093] Input layer: Current and voltage characteristics are used as inputs, and corresponding nodes are set up to receive preprocessed input data;

[0094] Hidden layer: Radial basis function: Using Gaussian kernel function:

[0095]

[0096] Center C j K-means clustering is used to group the input data, with the cluster centers serving as the base function centers.

[0097] Width σ j Take the mean of the maximum distance from each cluster sample to the center, or the mean of the distance between adjacent centers;

[0098] Output layer: Set 4 nodes, corresponding to whether the 4 parameters are out of bounds. Each node uses the Sigmoid activation function to map the output value to the (0, 1) interval, representing the probability of the parameter being out of bounds;

[0099] S400: Using the neural network model trained in step S300, input real-time power battery monitoring data to determine the state of the battery, and then analyze the aging phenomena that will occur in the battery in the future, and carry out corresponding control for different aging phenomena.

[0100] The real-time acquired detection data includes battery temperature, depth of charge / discharge, and charge / discharge rate. Based on fuzzy PID control and considering the coupling of battery control variables, the control variables are decoupled before control is implemented to accurately control battery aging. Considering the uncertainty and self-feedback characteristics of the controllable variables, fuzzy PID control is introduced to reasonably constrain the controllable variables of the battery, thereby increasing the flexibility and universality of aging detection and improving the accuracy of aging suppression control methods.

[0101] During charging, current control is a common control method. When an aging signal is detected, PID control can be used to slow down battery aging while ensuring sufficient charging speed.

[0102] Specific implementation scheme 2: The present invention provides a battery aging delay system based on a matched aging mode. The system has a program module corresponding to the above steps, and executes the steps in the above-mentioned battery aging delay method based on a matched aging mode when running.

[0103] The other combinations and connections in this implementation scheme are the same as in Specific Implementation Scheme 1.

[0104] Specific Implementation Scheme 3: The present invention provides a computer-readable storage medium storing a computer program configured to implement, when called by a processor, the steps of a battery aging delay method based on a matched aging mode.

[0105] The other combinations and connections in this implementation scheme are the same as in Specific Implementation Scheme 1.

[0106] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A battery aging delay method based on matched aging modes, characterized in that, Includes the following steps: S100. Conduct a cycle aging test on the battery to obtain the battery aging condition data. Use the condition data as the result of battery cycle aging. Based on the pseudo two-dimensional model, perform parameter identification to obtain the internal state of the lithium-ion battery. S200: Match the aging mode using the parameter identification results obtained in step S100. The aging mode includes SEI film growth, lithium dendrite formation, electrolyte decomposition, and active material loss. S300: The matching analysis results obtained in step S200 are used as training data and input into the radial basis neural network for training to obtain the trained neural network model. S400: Using the neural network model trained in step S300, input real-time power battery monitoring data to determine the state of the battery, and then analyze the aging phenomena that will occur in the battery in the future, and carry out corresponding control for different aging phenomena.

2. The battery aging delay method based on matched aging mode according to claim 1, characterized in that: In step S100, parameter identification using the pseudo-two-dimensional model includes, Charge conservation equation for solid particles: Where, φ s Represents the potential of a solid particle; σ eff Represents the effective solid particle conductivity; F represents the Faraday constant; j represents the ion flux; a s denoted by , where represents the surface area of ​​the solid particle; x represents the spatial coordinates of the internal potential change of the solid particle. The mass conservation equation for solid particles: Among them, c s Represents the lithium concentration in the solid phase (related to position r and time t); t represents time; r represents the radial coordinate of the particle; Represents the effective solid particle diffusion rate; Electrolyte charge conservation equation: Among them, κ eff Represents the effective electrolyte conductivity; φ e Represents electrolyte potential; R represents universal gas constant; T represents temperature; t + Represents the migration number; c e This represents the lithium concentration in the electrolyte phase; Electrolyte mass conservation equation: Where, ε p,n Represents porosity; Represents the effective electrolyte diffusivity; ε s The volume percentage of solid particles; The equation of lithium-ion motion between solid particles and electrolyte: Where, k p,n Represents the reaction rate constant; This represents the maximum concentration of lithium in the solid phase. η represents the lithium concentration at the solid surface; η represents the overpotential; and α represents the transfer coefficient.

3. The battery aging delay method based on matched aging mode according to claim 1, characterized in that: In step S200, For SEI film growth: effective electrolyte conductivity κ eff and effective electrolyte diffusion rate Changes have occurred; For lithium dendrite formation: effective electrolyte conductivity κ eff Effective electrolyte diffusion rate solid particle surface area a s Effective solid particle diffusion rate And effective solid particle conductivity σ eff Changes have occurred; For electrolyte decomposition: effective electrolyte diffusivity Changes have occurred; For active material loss: effective solid particle diffusion rate With the surface area a of solid particles s Things have changed.

4. The battery aging delay method based on matched aging modes according to claim 1, characterized in that: In step S400, the real-time acquired detection data includes battery temperature, depth of charge / discharge, and charge / discharge rate. Based on fuzzy PID control and considering the coupling of battery control quantities, the control quantities are decoupled before control to control battery aging.

5. The battery aging delay method based on matched aging mode according to claim 1, characterized in that: In step S400, when the battery is subjected to a cycle aging test, the corresponding external signal data of the battery is obtained by conducting cycle aging tests at different temperatures, and then input into the battery pseudo-two-dimensional model and parameter identification. The fitness function is the minimum root mean square of the model voltage and the measured voltage, and a genetic algorithm is used to identify the parameters.

6. A battery aging delay system based on matched aging modes, characterized in that: The system has a program module corresponding to the steps of any one of the claims 1-5 above, and executes the steps in the above-described battery aging delay method based on matching aging modes when running.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program configured to, when invoked by a processor, implement the steps of any one of claims 1-5, a battery aging delay method based on a matched aging mode.

Images