A Parameter Identification Method for Lithium Batteries Based on the Limited Memory Method with Double Memory Lengths

Abstract

The present invention provides a lithium battery parameter identification method based on a limited memory method with double memory lengths, comprising the following steps: Step 1): Obtain the terminal current and terminal voltage information of the lithium battery; Step 2): Classify the parameters in the circuit model according to the Thevenin equivalent circuit model and the characteristics of each parameter in the model; Step 3): Obtain the transfer function of the circuit according to the Thevenin equivalent circuit model, and obtain an autoregressive model with an external input after discretization; Step 4): Construct the calculation process of the limited memory method with double memory lengths, and use the limited memory method with double memory lengths to identify the parameters of the autoregressive model, and calculate the parameters of the equivalent circuit model. Through the lithium battery parameter identification method based on the limited memory method with double memory lengths of the present invention, the influence brought by different characteristics of model parameters in traditional lithium battery parameter identification algorithms can be solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of lithium - ion batteries, and particularly relates to a lithium - battery parameter identification method based on a limited - memory method with double memory lengths. Background Art

[0002] Lithium - ion batteries are currently widely used in electric vehicles, and their extensive use can effectively reduce the environmental pollution of automobiles. At the same time, as a complex non - linear system, the accurate modeling and model parameter identification of lithium batteries have always been research hotspots. Currently, the lithium - ion battery model generally adopts an equivalent circuit model, which is a variable - parameter system, and online parameter identification can effectively solve the problem of parameter time - variation.

[0003] However, the equivalent circuit model of lithium ions has the problem that model parameters have different time scales, which easily leads to differences in the identification effects of different model parameters when using the same sampling step. Specifically, when using a smaller sampling step, the parameters with slower - changing terminal voltages are prone to inaccurate identification. Similarly, when using a larger sampling step, the parameters with faster - changing terminal voltages are prone to inaccurate identification. In addition, most of the existing identification algorithms require the known OCV curve, which requires a large amount of test time and is also prone to errors in the initially used OCV curve due to reasons such as external temperature changes and long battery usage time. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies of the prior art and provide a lithium - battery parameter identification method based on a limited - memory method with double memory lengths, which can solve the influence brought by different characteristics of model parameters in traditional lithium - battery parameter identification algorithms.

[0005] The present invention provides a lithium - battery parameter identification method based on a limited - memory method with double memory lengths, including the following steps:

[0006] Step 1): Obtain the terminal current and terminal voltage information of the lithium battery;

[0007] Step 2): Classify the parameters in the circuit model according to the Thevenin equivalent circuit model and the characteristics of each parameter in the model;

[0008] Step 3): Obtain the transfer function of the circuit according to the Thevenin equivalent circuit model, and obtain an autoregressive model with an external input after discretization;

[0009] Step 4): Construct the calculation process of the limited - memory method with double memory lengths, and use the limited - memory method with double memory lengths to identify the parameters of the autoregressive model, and calculate the parameters of the equivalent circuit model.

[0010] Furthermore, the Thevenin equivalent circuit model described in step 2) includes 4 parameters, namely: ideal voltage source Uoc, resistors R0 and R1, and capacitor C1. Among them, Uoc represents the open-circuit voltage (OCV) of the battery, R0 is the internal resistance of the battery, and R1 and C1 are an RC link in the equivalent circuit model.

[0011] Furthermore, step 3) specifically includes the following steps:

[0012] Step 3.1): The transfer function obtained according to the Thevenin equivalent circuit model is:

[0013]

[0014] We let:

[0015]

[0016] The discretized form of the transfer function of the Thevenin equivalent circuit model is obtained as:

[0017]

[0018] where Ts is the sampling time of the system, and c1, c2, and c3 are coefficients composed of the parameters of the equivalent circuit model, which are respectively:

[0019]

[0020] where 1 = R1C1.

[0021] Furthermore, step 3) specifically includes the following steps: Step 3.2): The autoregressive model of the Thevenin equivalent circuit model obtained according to the discretized transfer function is:

[0022] u t (k) = c1u t (k - 1) + c2i t (k) + c3i t (k - 1) + u oc (k) - c1u oc (k - 1)

[0023] Also considering that uoc changes little within a unit sampling time, it can be considered that:

[0024] u oc (k) = u oc (k - 1)

[0025] Then the autoregressive model of the Thevenin equivalent circuit model can be simplified to:

[0026] u t (k) = c1u t (k - 1) + c2it (k) + c3i t (k - 1) + (1 - c1)u oc (k).

[0027] Furthermore, step 3) specifically includes the following steps: Step 3.3): Write the autoregressive model of the simplified Thevenin equivalent circuit model in matrix form as:

[0028]

[0029] where y(k) is the output of the model at time k, Φ(k) is the data vector of the model at time k, and θ(k) is the parameter vector of the model at time k.

[0030] Furthermore, step 4) specifically includes the following steps: Step 4.1): Given the initial conditions (0, 0) and P(0, 0), and given the small memory length L1 and the large memory length L2,

[0031]

[0032] Furthermore, step 4) specifically includes the following steps: Step 4.2): Use the following least-squares recursive algorithm to obtain the preliminary parameter estimates (0, L1 - 1), the covariance matrix P(0, L1 - 1), and the gain coefficient K(0, L1 - 1) as the initial state of the limited-memory recursive algorithm. At this time, the small memory length and the large memory length share a set of parameters,

[0033]

[0034] Step 4.2): When k runs to L1, take out the parameters (0, L1 - 1), P(0, L1 - 1), and K(0, L1 - 1) required for the small memory length, and start using the following formula for the limited-memory parameter identification algorithm with a memory length of L1. The large memory length continues with the least-squares recursive algorithm,

[0035]

[0036] Furthermore, step 4) specifically includes the following steps: Step 4.3): When k runs to L2, take out the parameters (0, L2 - 1), P(0, L2 - 1), and K(0, L2 - 1) required for the large memory length, and start using the following formula for the limited-memory parameter identification algorithm with a memory length of L2,

[0037]

[0038] Further, step 4) specifically includes the following steps: Step 4.4): Starting from time L2, the parameter identification algorithm with limited memory is performed simultaneously on the large memory length parameter and the small memory length parameter.

[0039] Further, substituting the autoregressive model of the simplified Thevenin equivalent circuit model into the limited memory method with double memory lengths to obtain c1, c2, c3, and c4, the parameter expressions of the Thevenin equivalent circuit model can be solved as follows:

[0040]

[0041] Among them, R0, R1, and Uoc are calculated from the small memory length parameter, and C1 is calculated from the large memory length parameter.

[0042] Through the lithium battery parameter identification method based on the limited memory method with double memory lengths of the present invention, the influence brought by different characteristics of model parameters in the traditional lithium battery parameter identification algorithm can be solved. Description of the Drawings

[0043] Figure 1 is the Thevenin equivalent circuit model.

[0044] Figure 2 is the algorithm structure diagram of the present invention. Detailed Embodiments

[0045] The following will clearly and completely describe the concept, specific structure, and technical effects generated by the present invention in combination with the embodiments and the drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The same reference numerals used throughout the drawings indicate the same or similar parts.

[0046] The specific structure of a multi-time scale step-by-step parameter identification method for lithium batteries provided by the present invention is as Figure 2 shown, and specifically includes the following steps:

[0047] Step 1): Real-time obtain the terminal current and terminal voltage information of the lithium battery. According to the application scenario of the present invention, it means using the sensors on the electric vehicle to measure the terminal current and terminal voltage of the power lithium battery in real time;

[0048] Step 2): According to the Thevenin equivalent circuit model and the characteristics of each parameter in the model, distinguish each parameter in the circuit model into different time scales;

[0049] Further, the Thevenin equivalent circuit model described in step 2) is as Figure 1As shown, it includes four parameters, namely: ideal voltage source Uoc, resistors R0 and R1, and capacitor C1. Among them, Uoc represents the open-circuit voltage (OCV) of the battery, R0 is the internal resistance of the battery, and R1 and C1 are an RC link in the equivalent circuit model.

[0050] According to the existing prior knowledge, R0 and R1 are considered to have a relatively fast response rate. C1 is considered to have a relatively slow response rate. This is easy to understand: R0 and R1 are resistors, and the voltage across their terminals can change abruptly. C1 is a capacitor, and the voltage across its terminals cannot change abruptly but can only change gradually. Uoc, as a DC voltage source, is also considered to have a relatively fast response rate.

[0051] Step 3): Obtain the transfer function of the circuit according to the Thevenin equivalent circuit model, and after discretization, obtain an autoregressive model with an external input;

[0052] Furthermore, step 3) specifically includes the following steps:

[0053] Step 3.1): According to Figure 1 the Thevenin equivalent circuit model, its transfer function is:

[0054]

[0055] We let:

[0056]

[0057] The discretized form of the transfer function of the Thevenin equivalent circuit model is:

[0058]

[0059] where Ts is the sampling time of the system, and c1, c2, and c3 are coefficients composed of the parameters of the equivalent circuit model, which are respectively:

[0060]

[0061] where τ = R1C1.

[0062] Step 3.2): According to the discretized transfer function, the autoregressive model of the Thevenin equivalent circuit model is:

[0063] u t (k) = c1u t (k - 1) + c2i t (k) + c3i t (k - 1) + u oc (k) - c1u oc (k - 1)

[0064] Also considering that uoc changes little within the unit sampling time, it can be considered that:

[0065] u oc (k) = u oc (k - 1)

[0066] Then the autoregressive model of the Thevenin equivalent circuit model can be simplified to:

[0067] u t (k) = c1u t (k - 1) + c2i t (k) + c3i t (k - 1) + (1 - c1)u oc (k)

[0068] Step 3.3): Write the autoregressive model of the simplified Thevenin equivalent circuit model in matrix form as:

[0069]

[0070] where y(k) is the output of the model at time k, Φ(k) is the data vector of the model at time k, and θ(k) is the parameter vector of the model at time k.

[0071] Step 4): As Figure 2 shown, construct the calculation process of the limited memory method with double memory lengths. And use the limited memory method with double memory lengths to identify the parameters of the autoregressive model and calculate the parameters of the equivalent circuit model;

[0072] Further, Step 4) specifically includes the following steps:

[0073] Step 4.1): Given the initial conditions (0, 0) and P(0, 0), and given the small memory length L1 and the large memory length L2.

[0074]

[0075] Step 4.2): Use the following least squares recursive algorithm to obtain the preliminary parameter estimates (0, L1 - 1), the covariance matrix P(0, L1 - 1), and the gain coefficient K(0, L1 - 1) as the initial state of the limited memory recursive algorithm. At this time, the small memory length and the large memory length share a set of parameters.

[0076]

[0077] Step 4.2): When k runs to L1, take out the parameters required for the short memory length (0, L1 - 1), P(0, L1 - 1), and K(0, L1 - 1), and start using the following formula to perform the limited memory parameter identification algorithm with a memory length of L1. The large memory length continues to perform the least squares recursive algorithm.

[0078]

[0079] Step 4.3): When k runs to L2, take out the parameters required for the large memory length (0, L2 - 1), P(0, L2 - 1), and K(0, L2 - 1), and start using the following formula to perform the limited memory parameter identification algorithm with a memory length of L2.

[0080]

[0081] Step 4.4): Starting from the L2 moment, both the large memory length parameters and the short memory length parameters perform the limited memory parameter identification algorithm.

[0082] Step 4.5): Substitute the autoregressive model of the simplified Thevenin equivalent circuit model into the limited memory method with double memory lengths to obtain c1, c2, c3, and c4, and then the parameter expressions of the Thevenin equivalent circuit model can be solved as follows:

[0083]

[0084] Among them, R0, R1, and Uoc are calculated from the short memory length parameters, and C1 is calculated from the large memory length parameters.

[0085] The above embodiments are only used to illustrate the technical concept and features of the present invention. The purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and it cannot limit the protection scope of the present invention. All equivalent changes and modifications made to the scope of the claims of the present invention shall fall within the scope covered by the claims of the present invention.

[0086] As mentioned above, it is only the preferred embodiment of the present invention. The present invention is not limited to the above implementation manners. As long as it achieves the technical effects of the present invention by the same means, it should fall within the protection scope of the present invention. Within the protection scope of the present invention, various different modifications and changes can be made to its technical solutions and / or implementation manners.

Claims

1. A method for identifying lithium battery parameters based on a limited memory method with double memory lengths, characterized in that, It includes the following steps: Step 1): Obtain the terminal current and terminal voltage information of the lithium battery; Step 2): Classify the parameters in the circuit model according to the Thevenin equivalent circuit model and the characteristics of each parameter in the model; Step 3): Obtain the transfer function of the circuit according to the Thevenin equivalent circuit model, and obtain the autoregressive model with external input after discretization; Step 4): Construct the calculation process of the limited memory method with double memory lengths, and use the limited memory method with double memory lengths to identify the parameters of the autoregressive model, and calculate the parameters of the equivalent circuit model, where specifically includes: Step 4.1): Given the initial conditions and P(0, 0), and given a small memory length L1 and a large memory length L2 Step 4.2): Obtain preliminary parameter estimates using the following least squares recursive algorithm , covariance matrix and gain coefficient as the initial state of the bounded memory recursive algorithm. At this time, the small memory length and the large memory length share a set of parameters Step 4.3): When k runs to L1, the parameters required for the short memory length , and are taken out, and the limited memory parameter identification algorithm with a memory length of L1 starts to be used with the following formula. The least squares recursive algorithm continues for the long memory length. ； Step 4.4): When running to L2, take out the parameters required for the large memory length , and and start to use the following formula to perform the limited memory parameter identification algorithm with a memory length of L2. ； Step 4.5): Starting from time L2, perform the parameter identification algorithm of limited memory for both the large memory length parameter and the small memory length parameter; Step 4.6): Substitute the autoregressive model of the simplified Thevenin equivalent circuit model into the limited memory method with double memory lengths to obtain c1, c2, c3, and c4, and then the parameter expression of the Thevenin equivalent circuit model can be solved as: Among them, R0, R1, and U oc are calculated from the small memory length parameter, and C1 is calculated from the large memory length parameter.

2. The lithium battery parameter identification method based on the limited memory method with double memory lengths as described in claim 1, wherein The Thevenin equivalent circuit model described in step 2) includes 4 parameters, namely: ideal voltage source U oc , resistors R0 and R1, and capacitor C1. Among them, U oc represents the open circuit voltage (OCV) of the battery, R0 is the internal resistance of the battery, and R1 and C1 are an RC link in the equivalent circuit model.

3. The lithium battery parameter identification method based on the limited memory method with double memory lengths according to claim 1, characterized in that, Step 3) specifically includes the following steps: Step 3.1): According to the Thevenin equivalent circuit model, the transfer function is obtained as: Let: The discretized form of the transfer function of the Thevenin equivalent circuit model is obtained as: where is the sampling time of the system, and c1, c2, and c3 are coefficients composed of equivalent circuit model parameters, which are respectively: Among them, = R1C1.

4. The lithium battery parameter identification method based on the limited memory method with double memory lengths as described in claim 3, characterized in that, Step 3) specifically includes the following steps: Step 3.2): According to the discretized transfer function, the autoregressive model of the Thevenin equivalent circuit model is obtained as: Also considering that u oc changes little within a unit sampling time, and u oc represents the open-circuit voltage of the battery, so it can be considered that: Then the autoregressive model of the Thevenin equivalent circuit model can be simplified to: 。 5. The lithium battery parameter identification method based on the limited memory method with double memory lengths as described in claim 4, characterized in that, Step 3) specifically includes the following steps: Step 3.3): Write the autoregressive model of the simplified Thevenin equivalent circuit model in matrix form as: where \(y(k)\) is the output of the model at time step \(k\), is the data vector of the model at time step \(k\), is the parameter vector of the model at time step \(k\).

Images