A method for estimating key characteristic parameters of lithium batteries used for brake energy recovery in electric mining dump trucks

Abstract

The present invention discloses a method for estimating the key parameters of a lithium battery used for braking energy recovery in an electric-driven dump truck for mining. The estimation method uses a fractional-order equivalent circuit model (FOM) and integrates a dual adaptive unscented Kalman algorithm (DAUKF) to achieve multi-time-scale cyclic estimation of the key characteristic parameters of the lithium battery in the braking energy recovery system of the electric-driven dump truck for mining, providing an accurate data basis for the efficient and safe recovery of braking energy. The present invention improves the defects of existing methods for estimating characteristic parameters of lithium batteries, thereby achieving rapid and accurate estimation of characteristic parameters of lithium batteries under extreme working conditions such as frequent charging and discharging, and large current shocks. The safety and stability of the braking energy recovery system of electric-driven dump trucks for mining are improved. The present invention can solve the technical problem that existing methods for estimating key characteristic parameters of lithium batteries are mainly aimed at lithium batteries in ordinary working environments, but are not suitable for estimating key parameters of lithium batteries for braking energy recovery in electric-driven dump trucks for mining.

Description

Technical Field

[0001] The present invention belongs to the fields of novel energy storage technology and high-end intelligent manufacturing technology, and more specifically, relates to a method for estimating key characteristic parameters (KBP) of a lithium battery for braking energy recovery in an electric-driven mining dump truck. Background Art

[0002] Electric-powered dump trucks for mining are key equipment in global resource extraction and large-scale infrastructure construction. They undertake heavy transportation tasks in large-scale projects such as my country's large open-pit coal mines and the Three Gorges Dam on the Yangtze River. These trucks are often used on upper and lower sections of broken roads, resulting in extremely high braking energy consumption, severe wear, and harsh operating conditions. To prevent the loss of braking energy as heat, a hybrid energy storage system using supercapacitors and lithium batteries has been proposed to recover the braking energy of mining vehicles. Lithium batteries are often chosen for this energy storage system due to their significant advantages, such as high energy density, low self-discharge rate, and long life. Key characteristic parameters of lithium batteries include state of charge (SOC) and state of health (SOH). Accurate estimation of these parameters is crucial for the safe and stable operation of lithium batteries, facilitates the accurate allocation of energy proportions in the hybrid energy storage system, and is therefore crucial for the safe operation of the entire brake energy recovery system.

[0003] At present, the estimation methods for the key characteristic parameters of lithium batteries for brake energy recovery are mainly divided into three categories: the first category is the estimation method based on experiments, which includes the ampere-hour integration method, the open circuit voltage method, etc. The ampere-hour integration method is based on the integration of the current change and time when the lithium battery is working to express the change in charge, so as to estimate the current SOC of the lithium battery and then use the algorithm to estimate the SOH. The open circuit voltage method is based on the correlation between the open circuit voltage (OCV) of the battery in a static state and the SOC to estimate the SOC and then use the algorithm to estimate the SOH; the second category is the model-based estimation method, which constructs relevant lithium battery models such as equivalent circuit models and electrochemical models, identifies relevant parameters of the model, and realizes the estimation of the key characteristic parameters of the lithium battery; the third category is based on data-driven estimation methods, which rely on the mapping relationship between the input and output of the battery system and train a large number of data sets to estimate the key characteristic parameters of the battery.

[0004] However, the above-mentioned estimation methods of key characteristic parameters of lithium batteries all have some defects that cannot be ignored: First, although the ampere-hour integration method is simple to calculate, it has a large cumulative error and relies on the initial SOC value for estimation, but the initial SOC value is difficult to obtain accurately, which leads to low estimation accuracy; Second, the open circuit voltage method is simple to operate, but the battery must be kept stationary for more than ten hours when measuring the open circuit voltage. This method cannot meet the actual working conditions of the battery and cannot estimate the key characteristic parameters of the lithium battery in real time; Third, for the model-based estimation method, the model is divided into an internal characteristic model and an external characteristic model. Although the internal characteristic model (mainly the electrochemical model) can better fit the internal chemical reaction of the lithium battery, the calculation is very complicated, and the external characteristic model (mainly the internal resistance model) is not suitable for the internal chemical reaction of the lithium battery. , Thevenin model, PNGV model, etc.) If a simple model is used, the estimation accuracy will be low, and if a complex model is used, it will be difficult to build and the parameter identification will be difficult; Fourth, the data-driven estimation method requires a large amount of data to be obtained in advance for training. Over-reliance on the data set without considering the battery model will lead to overfitting; Fifth, the existing estimation methods for the key characteristic parameters of these lithium batteries are mainly aimed at lithium batteries in ordinary working environments, but are not suitable for the key parameter estimation of lithium batteries for brake energy recovery of electric-driven dump trucks for mining (this is because electric-driven dump trucks for mining mostly work on uphill and downhill sections, are frequently charged and discharged, and are large in size. When performing brake energy recovery, lithium batteries usually face the impact of large currents, extremely high braking energy consumption, and harsh working environments, facing extreme temperature differences of -20°C to 40°C and dusty environments). Summary of the Invention

[0005] In response to the above-mentioned deficiencies or improvement needs in the prior art, the present invention provides a method for estimating key parameters of a lithium battery for brake energy recovery in an electric-driven dump truck for mining. The method aims to address the technical problems that the existing ampere-hour integration method has a large cumulative error and relies on an initial SOC value for estimation, but the initial SOC value is difficult to accurately obtain, resulting in low estimation accuracy. The existing open-circuit voltage method requires the battery to be stationary for more than ten hours when measuring the open-circuit voltage, which cannot meet the actual working conditions of the battery and cannot estimate the key characteristic parameters of the lithium battery for brake energy recovery in real time. The existing model-based estimation method is very complex in calculation when using an intrinsic characteristic model, has low estimation accuracy when using an extrinsic characteristic model, and is difficult to identify parameters. The existing data-driven estimation method requires a large amount of data to be acquired in advance for training, and over-reliance on the data set without considering the battery model will lead to overfitting. The existing method for estimating key characteristic parameters of a lithium battery for brake energy recovery is mainly targeted at lithium batteries for brake energy recovery in ordinary working environments, but is not suitable for estimating key parameters of a lithium battery for brake energy recovery in electric-driven dump trucks for mining.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for estimating key characteristic parameters of a lithium battery for braking energy recovery in an electric mining dump truck is provided, comprising the following steps:

[0007] (1) Simulate and model the lithium battery for braking energy recovery using the SIMULINK tool in MATLAB software to obtain a fractional-order equivalent circuit model;

[0008] (2) The fractional-order equivalent circuit model obtained in step (1) is processed using Kirchhoff's current law and Kirchhoff's voltage law, respectively, to obtain the electrical principle equation and output equation of the fractional-order equivalent circuit model.

[0009] (3) The Greenwald-Letnikoff formula is used to process the electrical principle equation and output equation of the fractional-order equivalent circuit model established in step (2) to obtain the state space equation and observation equation defined in fractional order.

[0010] (4) Discretize the state equation and observation equation defined by the fractional order obtained in step (3) to obtain the discretized state equation and observation equation respectively.

[0011] (5) using an online parameter identification method to process the discretized state equation and observation equation obtained in step (4) respectively to obtain parameter data in the fractional-order equivalent circuit;

[0012] (6) Input the parameter data obtained by online parameter identification in step (5) into the fractional-order adaptive unscented Kalman algorithm FAUKF for iterative calculation to obtain the SOC at time t t Estimated value and ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t 0,t , the algorithm flow chart is as follows Figure 4 As shown:

[0013] (7) Determine whether the number of iterations in step (6) is greater than 60. If so, proceed to step (8); otherwise, return to step (6).

[0014] (8) Obtain multiple SOC estimation values ​​of the lithium battery for braking energy recovery obtained by 60 iterations and the ohmic internal resistance parameter R of the fractional-order equivalent circuit model obtained by the 60th iteration 0,t .

[0015] (9) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t , and obtain the estimated state of health (SOH) value of the lithium battery used for braking energy recovery through the FAUKF algorithm.

[0016] (10) Displaying the multiple SOC estimation values ​​of the lithium battery for braking energy recovery obtained in step (8) and the SOH estimation value obtained in step (9) to the user of the electric drive dump truck for mining.

[0017] Preferably, the electrical principle equation of the fractional-order equivalent circuit model is:

[0018]

[0019] Wherein, U1(t) and U2(t) represent the voltage drops of the two RC loops at time t in the fractional-order equivalent circuit model at the current moment; R1 represents the electrochemical polarization resistance in the fractional-order equivalent circuit model; CPE1 represents the electrochemical polarization capacitance in the fractional-order equivalent circuit model; R2 is the concentration polarization resistance in the fractional-order equivalent circuit model; CPE2 is the concentration polarization capacitance in the fractional-order equivalent circuit model; I t is the operating voltage of the fractional-order equivalent circuit model; n1 and n2 are the fractional orders of the electrochemical polarization capacitance and concentration polarization capacitance in the fractional-order equivalent circuit model, and their values ​​range from 0 to 1; The SOC expression for the state of charge of the lithium battery for braking energy recovery at the next moment, SOC t is the SOC and SOC of the equivalent circuit model at the current moment t+1 is the SOC of the equivalent circuit model at the next moment; T s is the system sampling period of the equivalent circuit model; Q n is the battery capacity of the equivalent circuit model.

[0020] The output equation of the fractional-order equivalent circuit model is:

[0021] U0(t)=U oc (SOC)-R0I(t)-U1(t)-U2(t)

[0022] Where U0(t) represents the terminal voltage of the fractional-order equivalent circuit model at the current moment, U oc (SOC) represents the open-circuit voltage of the fractional-order equivalent circuit model, R0 represents the ohmic internal resistance of the fractional-order equivalent circuit model, U1(t) and U2(t) are the two RC loop voltage drops of the fractional-order equivalent circuit model at the current moment.

[0023] Preferably, specifically, the fractional derivative definition formula is:

[0024]

[0025] Among them D n Fractional differential operators representing the state space equations and observation equations; D nf(t) represents the calculus of f(t); n is the order of the fractional-order equivalent circuit model, ranging from [0,1]; n! is the factorial of the order of the fractional-order equivalent circuit model; T s is the system sampling period of the fractional-order equivalent circuit model, ranging from [0, 1], preferably 1s; L is the memory length of the fractional-order equivalent circuit model, and L = 60; i is the number of sampling times; is the coefficient of the Newton binomial.

[0026] The state space equation and observation equation defined by the fractional order in step (3) are shown in the following formulas:

[0027]

[0028] Among them, [U 1,k U 2,k SOC k ] T is the input of the state-space equation, which represents the voltage drop of loop 1, the voltage drop of loop 2 and the state of charge of the fractional-order equivalent circuit model at the current moment; [U 1,k+1 U 2,k+1 SOC k+1 ] T represents the voltage drop of loop 1, the voltage drop of loop 2, and the state of charge of the fractional-order equivalent circuit model at the next moment; U oc (SOC k ) is the open circuit voltage as a function of the state of charge; I k is the working current of the fractional-order equivalent circuit model at the current moment; ω k is the system noise of the fractional-order equivalent circuit model at time k, which is caused by the uncertainty of the fractional-order equivalent circuit model; k is the measurement noise of the fractional-order equivalent circuit model at time k, which is caused by the measurement error of the fractional-order equivalent circuit model.

[0029] Preferably, the state equation and observation equation after discretization in step (4) are as shown in the following formulas:

[0030]

[0031] Among them, x k is the system input of the fractional-order equivalent circuit model at time k, and x k-1 is the system input of the fractional-order equivalent circuit model at time k-1; I k-1 is the operating current of the fractional-order equivalent circuit model at time k-1; ω k-1 is the system noise of the fractional-order equivalent circuit model at time k-1; y kis the system output of the fractional-order equivalent circuit model at time k; C = [-1 -1 0]; υ k is the measurement noise of the fractional-order equivalent circuit model at time k.

[0032] Preferably, the online parameter identification method in step (5) adopts a recursive least squares method based on minimum forgetting factor combined with a genetic algorithm, wherein the data to be identified of the fractional-order equivalent circuit includes the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2, and fractional orders n1 and n2 of the fractional-order equivalent circuit model.

[0033] Preferably, step (5) specifically includes the following sub-steps:

[0034] (5-1) Perform Laplace transform on the discrete state equation and observation equation obtained in step (4) to obtain the frequency domain expression of the fractional-order model, as shown in formula (7).

[0035]

[0036] Among them, Y(k) is the transfer function of the fractional-order equivalent circuit model in the time domain; Y(s) is the transfer function of the fractional-order equivalent circuit model in the frequency domain; s is the independent variable of the transfer function, which is equivalent to the time state in the frequency domain; U oc (s) is the open-circuit voltage of the fractional-order equivalent circuit model in the frequency domain; U o (s) is the terminal voltage of the fractional-order equivalent circuit model in the frequency domain; I(s) is the operating current of the fractional-order equivalent circuit model in the frequency domain.

[0037] (5-2) Use the bilinear transformation method to process the frequency domain expression of the fractional-order model obtained in step (5-1) to obtain the transformation form G(z) of the fractional-order model in the z domain:

[0038]

[0039] Among them, a1, a2, a3, a4, and a5 are the parameter identification data of the fractional-order equivalent circuit model, and the mathematical relationship between them and the data to be identified of the fractional-order equivalent circuit model is:

[0040] By utilizing the mathematical relationship between the parameter identification data of the fractional-order equivalent circuit model and the data to be identified of the fractional-order equivalent circuit model, the data to be identified of the parameters of the fractional-order equivalent circuit model can be obtained.

[0041] (5-3) Simplify the transformed form of the fractional-order model obtained in step (5-2) in the z-domain to obtain the recursive equation Y(k) of the fractional-order equivalent circuit model.

[0042] Y(k)=a1Y(k-1)+a2Y(k-2)+a3I(k)+a4I(k-1)+a5I(k-2)

[0043] (5-4) Initialize the parameter identification column vector θ(k) and covariance P(k) at time k obtained in step (5-3) to obtain the parameter identification column vector θ(0) and covariance P(0) at time k=0 after initialization.

[0044] (5-5) Determine whether k is equal to 60. If so, go to step (5-10), otherwise go to step (5-6).

[0045] (5-6) Use the initialized parameter identification column vector θ(0) and covariance P(0) obtained in step (5-5) to obtain the gain matrix K(k):

[0046]

[0047] Where K(k) is the estimated gain at time k; P(k-1) is the covariance at time k-1; λ is the forgetting factor, which ranges from [0.90, 1.00]. Considering the application scenarios of the present invention, 0.955 is preferred.

[0048] (5-7) The parameter identification column vector θ(k) in the recursive equation at time k is updated according to the gain matrix obtained in step (5-6) to obtain the updated parameter identification column vector θ(k) in the recursive equation at time k.

[0049]

[0050] Among them, e(k) is the sampling error at time k.

[0051] (5-8) Obtain the parameter data to be identified of the fractional-order equivalent circuit model based on the parameter identification column vector θ(k) in the updated k-time recursive equation obtained in step (5-7).

[0052] (5-9) The covariance P(k) is updated according to the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) to obtain an updated covariance P(k), and then the process returns to step (5-5):

[0053]

[0054] Wherein, P(k) is the updated covariance at time k; I is a 5-order scalar matrix, and the value range of the elements on the diagonal is [10e5, 10e6], preferably 10e6.

[0055] (5-10) Binary encoding the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) and the fractional orders n1 and n2 to obtain a gene population expressed in a coded form, wherein n1 and n2 are both 1 at the kth moment;

[0056] (5-11) Obtaining the population fitness based on the gene population expressed in coded form obtained in step (5-10):

[0057]

[0058] Where J(i) is the fitness of the population when the population size is i, and i∈[1, N]; N is the population size, preferably N=100; A(i) is the predicted output value U in the fractional-order equivalent circuit model y (y) and the actual measured output value U c The sum of squares of the differences between (y), A max is the optimization target of the genetic algorithm. In the present invention, A max =0.005; U y (j) is the predicted output value at the jth iteration; U c (j) is the actual measured output value at the jth iteration, and j∈[1,M]; M is the number of iterations, preferably M=60;.

[0059] (5-12) According to the population fitness obtained in step (5-11), the population is subjected to hybridization and mutation processing to obtain a processed population:

[0060]

[0061] Among them, P z (i) is the hybridization probability of the population when the population size is i; K z is the hybridization factor of the population, and its value range is (0.6, 0.9), preferably 0.7; P b (i) is the probability of population mutation when the population size is i; K b is the variation factor, and its value range is [0.01, 0.1], preferably 0.05; A ave is the average fitness in the population.

[0062] (5-13) Decode the population processed in step (5-12) to obtain a decoded population, which is the parameter data to be identified of the fractional-order equivalent circuit model and the fractional orders n1 and n2.

[0063] (5-14) Determine whether the number of iterations j is greater than 60. If so, proceed to step (5-15). Otherwise, set j=j+1 and return to step (5-11).

[0064] (5-15) Output the parameter data in the fractional-order equivalent circuit, including the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2 of the fractional-order equivalent circuit model, and the fractional orders n1 and n2.

[0065] Preferably, step (5) includes the following sub-steps:

[0066] (9-1) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t Obtain the discretized state-space equations and observation equations for estimating the ohmic internal resistance:

[0067] R 0,t+1 =R 0,t +r t

[0068] U 0,t =U oc,t (SOC)-R 0,t IU 1,t -U 2,t +t t

[0069] Among them, R 0,t+1 is the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1; R 0,t is the ohmic internal resistance of the fractional-order equivalent circuit model at time t; r t is the state noise of the state space equation estimating the ohmic internal resistance at time t; t is the observation noise of the observation equation for estimating the ohmic internal resistance at time t.

[0070] (9-2) Initialize the discretized state space equation and observation equation for estimating the ohmic internal resistance obtained in step (9-1).

[0071] Specifically, this step is to initialize the ohmic internal resistance of the fractional-order equivalent circuit model to R0 obtained by parameter identification in step (5); and to initialize the initial ohmic internal resistance covariance of the fractional-order equivalent circuit model to 0.

[0072] (9-3) Perform untraceable transformation (UT) on the discrete state space equation and observation equation initialized in step (9-2) to obtain multiple sigma points.

[0073] (9-4) Apply the multiple sigma points obtained in step (9-3) to the fractional-order equivalent circuit model obtained in step (1) to obtain the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t Variance prediction value P of the state space equation x , the mean weight corresponding to each sigma point and the covariance weights for the discrete state space equations

[0074] (9-5) The mean weight corresponding to the sigma point obtained in step (9-4) Covariance Weights for Discrete State-Space Equations Get the observation value of the observation equation at time t+1 and the observed variance predicted value P y .

[0075] (9-6) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t obtained according to step (9-4) The variance prediction value P of the discrete state space equation x , the mean weight corresponding to the sigma point Covariance Weights for Discrete State-Space Equations And the observation value of the observation equation at time t+1 obtained in step (9-5) and the observed variance predicted value P y , get the input-output cross covariance matrix P xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t .

[0076] (9-7) The input-output cross covariance matrix P obtained according to step (9-6) xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t , obtain the noise covariance Q of the discrete state space equation at time t+1 t+1 and the observation noise covariance R t+1 .

[0077] (9-8) The noise covariance Q of the discrete state space equation at time t+1 obtained using step (9-7) t+1 and the observation noise covariance R t+1 Get the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1

[0078] (9-9) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1 obtained in step (9-8) Get the estimated SOH value SOH of the lithium battery for braking energy recovery at time t+1 t+1 .

[0079] Preferably, the UT transformation process of step (9-3) is:

[0080]

[0081] in, is the mth sigma point at time t, representing the mathematical expectation of the state vector at time t; is the initial predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model; is the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t; m is the dimension of the state variable of the discrete state space equation after initialization in step (9-2); M is the length of the state vector of the discrete state space equation after initialization in step (9-2), and the value is determined by the number of state vectors. In the present invention, M is 3; γ is the scale parameter of the untraceable transformation; α is the divergence factor of the untraceable transformation, which is used to determine the sampling distribution of the sigma point, and the value range is 0.0001-1, preferably 0.005; β is the adjustment parameter of the untraceable transformation, which is used to provide other constraints on the sampling point, and the value is 0.

[0082] Step (9-4) uses the following calculation formula:

[0083]

[0084] Where χ is the weight coefficient, χ = 2; Q t is the noise covariance of the discrete state space equation at time t.

[0085] Preferably, step (9-5) updates the observation value and the observed variance predicted value P y The process is as follows:

[0086]

[0087] in, is the observation value of the updated observation equation at time t+1; y t is the actual observation value of the observation equation at time t; R t is the observation noise covariance of the observation equation at time t.

[0088] The update equation of step (9-6) is:

[0089]

[0090] in, is the transpose of the input-output cross-covariance matrix.

[0091] The implementation process of step (9-7) is shown below:

[0092]

[0093] Among them, G k is the adaptive variance of the moving window method; Z is the adaptive moving window.

[0094] The calculation formula for the predicted SOH value of the lithium battery for braking energy recovery at time t+1 in step (9-9) is as follows:

[0095]

[0096] Among them, SOH t+1 is the estimated SOH value of the lithium battery for braking energy recovery at time t+1; R N is the maximum available resistance of the lithium battery for braking energy recovery, is the rated ohmic internal resistance of the lithium battery for braking energy recovery; R e It is a healthy termination resistor for lithium batteries used for braking energy recovery. It is usually stipulated that if the capacity of lithium batteries used for braking energy recovery is less than 80% of the rated capacity, the battery must be scrapped. Therefore, R e The size of is the ohmic internal resistance of the lithium battery for braking energy recovery when the capacity of the lithium battery for braking energy recovery is reduced to 80% of the rated capacity.

[0097] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0098] (1) Since the present invention adopts steps (1) to (4) to establish a fractional-order equivalent circuit model, it can better fit the nonlinear system of the lithium battery for braking energy recovery in the braking energy recovery system compared to the direct and simple calculation of the ampere-hour integration method;

[0099] (2) Since the present invention adopts step (5), once the electric drive dump truck for mining starts to run uninterruptedly, the lithium battery for braking energy recovery in the braking energy recovery system is in a rapidly changing nonlinear time-varying system. The recursive least squares method based on the minimum forgetting factor can reduce the influence of old data, track the changes of the battery system more quickly, and the parameter identification is more in line with the system operation characteristics. At the same time, the genetic algorithm is used to identify the fractional order, avoiding the problem that the recursive least squares method based on the minimum forgetting factor cannot identify the fractional order online;

[0100] (3) Since the present invention adopts steps (6) to (8), the state of charge of the lithium battery for braking energy recovery is estimated in real time on a short time scale, which is consistent with the working conditions of the lithium battery for braking energy recovery and ensures the safe and stable operation of the lithium battery for braking energy recovery;

[0101] (4) Since the present invention adopts step (9), and since the health state of the lithium battery for braking energy recovery does not change significantly, the charge state of the lithium battery for braking energy recovery is estimated 60 times, and then the health state of the lithium battery for braking energy recovery is estimated once over a long time scale, thereby reducing the number of calculations while ensuring the accuracy of the estimation;

[0102] (5) Since the present invention adopts steps (6) to (9), the key characteristic parameters SOC and SOH of the lithium battery for braking energy recovery are jointly estimated in long-short time scales through two FAUKF algorithms, and the estimated result of SOC is used as the input of SOH estimation, which reduces calculation redundancy. The SOC and SOH measurement data are updated and corrected with each other, and real-time monitoring can be performed under harsh operating conditions to ensure safe and stable operation of the system.

[0103] (6) Since the present invention adopts step (10), the user can pay attention to the operating status of the lithium battery for braking energy recovery in real time, promptly perceive battery failure and replace aging batteries, thereby ensuring driving safety and improving economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1 This is a flow chart of a method for estimating key parameters of a lithium battery for braking energy recovery in an electric mining dump truck according to the present invention;

[0105] Figure 2 is a schematic diagram of a fractional-order equivalent circuit model of the present invention;

[0106] Figure 3 is a detailed flow chart of step (5) in the method of the present invention;

[0107] Figure 4 is a detailed flow chart of step (6) in the method of the present invention;

[0108] Figure 5 It is a detailed flow chart of step (9) in the method of the present invention. DETAILED DESCRIPTION

[0109] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0110] The present invention provides a method for estimating key parameters of a lithium battery used for braking energy recovery in an electric-drive dump truck for mining. The method uses a fractional order model (FOM) and integrates a double adaptive unscented Kalman filter (DAUKF) to achieve multi-time-scale cyclic estimation of key characteristic parameters of the lithium battery used for braking energy recovery in the braking energy recovery system of the electric-drive dump truck for mining, thereby ensuring safe and stable operation of the braking energy recovery system of the electric-drive dump truck for mining.

[0111] like Figure 1 As shown, the present invention provides a method for estimating key characteristic parameters of a lithium battery for braking energy recovery of an electric drive dump truck for mining, comprising the following steps:

[0112] (1) The SIMULINK tool in MATLAB software is used to simulate and model the lithium battery for braking energy recovery to obtain a fractional-order equivalent circuit model, such as Figure 2 shown.

[0113] Specifically, the lithium battery for braking energy recovery of the present invention uses a 6-DG-190 manganese acid lithium battery for braking energy recovery, and its specific parameter indicators are shown in the following Table 1:

[0114] Table 1 Parameters of lithium batteries for manganese acid brake energy recovery

[0115]

[0116]

[0117] (2) The fractional-order equivalent circuit model obtained in step (1) is processed using Kirchhoff's current law and Kirchhoff's voltage law, respectively, to obtain the electrical principle equation and output equation of the fractional-order equivalent circuit model.

[0118] Specifically, the electrical principle equation of the fractional-order equivalent circuit model is shown in formula (1):

[0119]

[0120] Wherein, U1(t) and U2(t) represent the voltage drops of the two RC loops at time t in the fractional-order equivalent circuit model, respectively; R1 represents the electrochemical polarization resistance in the fractional-order equivalent circuit model; CPE1 represents the electrochemical polarization capacitance in the fractional-order equivalent circuit model; R2 is the concentration polarization resistance in the fractional-order equivalent circuit model; CPE2 is the concentration polarization capacitance in the fractional-order equivalent circuit model; It is the operating voltage of the fractional-order equivalent circuit model; n1 and n2 are the fractional orders of the electrochemical polarization capacitance and concentration polarization capacitance in the fractional-order equivalent circuit model, and their values ​​range from 0 to 1; The SOC is the state of charge (SOC) expression of the lithium battery used for braking energy recovery. t is the SOC and SOC of the equivalent circuit model at the current moment t+1 is the SOC of the equivalent circuit model at the next moment; T s is the system sampling period of the equivalent circuit model; Q n is the battery capacity of the equivalent circuit model.

[0121] The output equation of the fractional-order equivalent circuit model is shown in formula (2):

[0122] U0(t)=U oc (SOC)-R0I(t)-U1(t)-U2(t) (2)

[0123] Where U0(t) represents the terminal voltage of the fractional-order equivalent circuit model, U oc (SOC) represents the open-circuit voltage of the fractional-order equivalent circuit model, R0 represents the ohmic internal resistance of the fractional-order equivalent circuit model, U1(t) and U2(t) are the two RC loop voltage drops of the fractional-order equivalent circuit model.

[0124] (3) The Greenwald-Letnikov formula (GL fractional-order derivative definition formula) is used to process the electrical principle equation and output equation of the fractional-order equivalent circuit model established in step (2) to obtain the state space equation and observation equation defined in fractional order.

[0125] Specifically, the definition formula of fractional derivative is shown in formula (3):

[0126]

[0127] Among them D n Fractional differential operators representing the state space equations and observation equations; D n f(t) represents the calculus of f(t); n is the order of the fractional-order equivalent circuit model, ranging from [0,1]; n! is the factorial of the order of the fractional-order equivalent circuit model; T s is the system sampling period of the fractional-order equivalent circuit model, ranging from [0, 1], preferably 1s; L is the memory length of the fractional-order equivalent circuit model, and L = 60; i is the number of sampling times; is the coefficient of the Newton binomial.

[0128] The state space equation and observation equation defined in fractional order in this step are shown in the following formulas (4) and (5), respectively:

[0129]

[0130] Among them, [U 1,k U 2,k SOC k ] T is the input of the state-space equation, which represents the voltage drop of loop 1, the voltage drop of loop 2 and the state of charge of the fractional-order equivalent circuit model at the current moment; [U 1,k+1 U 2,k+1 SOC k+1 ] T represents the voltage drop of loop 1, the voltage drop of loop 2, and the state of charge of the fractional-order equivalent circuit model at the next moment; U oc (SOC k ) is the open circuit voltage as a function of the state of charge; I k is the working current of the fractional-order equivalent circuit model at the current moment; ω k is the system noise of the fractional-order equivalent circuit model at time k, which is caused by the uncertainty of the fractional-order equivalent circuit model; k is the measurement noise of the fractional-order equivalent circuit model at time k, which is caused by the measurement error of the fractional-order equivalent circuit model.

[0131] (4) Discretize the state equation and observation equation defined by the fractional order obtained in step (3) to obtain the discretized state equation and observation equation respectively.

[0132] Specifically, the state equation and observation equation after discretization are shown in the following formula (6):

[0133]

[0134] Among them, x k is the system input of the fractional-order equivalent circuit model at time k, and x k-1 is the system input of the fractional-order equivalent circuit model at time k-1; I k-1 is the operating current of the fractional-order equivalent circuit model at time k-1; ω k-1 is the system noise of the fractional-order equivalent circuit model at time k-1; y k is the system output of the fractional-order equivalent circuit model at time k; C = [-1 -1 0]; υ k is the measurement noise of the fractional-order equivalent circuit model at time k.

[0135] The advantage of the above steps (1) to (4) is that a fractional-order equivalent circuit model is established, which can better fit the nonlinear system of the lithium battery for braking energy recovery than other models.

[0136] (5) Use the online parameter identification method to process the discretized state equation and observation equation obtained in step (4) respectively to obtain the parameter data in the fractional-order equivalent circuit, such as Figure 3 shown.

[0137] Specifically, the online parameter identification method adopts the recursive least squares method based on the minimum forgetting factor combined with the genetic algorithm, where the data to be identified of the fractional-order equivalent circuit includes the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2, and fractional orders n1 and n2 of the fractional-order equivalent circuit model.

[0138] This step specifically includes the following sub-steps:

[0139] (5-1) Perform Laplace transform on the discrete state equation and observation equation obtained in step (4) to obtain the frequency domain expression of the fractional-order model, as shown in formula (7).

[0140]

[0141] Among them, Y(k) is the transfer function of the fractional-order equivalent circuit model in the time domain; Y(s) is the transfer function of the fractional-order equivalent circuit model in the frequency domain; s is the independent variable of the transfer function, which is equivalent to the time state in the frequency domain; U oc (s) is the open-circuit voltage of the fractional-order equivalent circuit model in the frequency domain; U o (s) is the terminal voltage of the fractional-order equivalent circuit model in the frequency domain; I(s) is the operating current of the fractional-order equivalent circuit model in the frequency domain.

[0142] (5-2) Use the bilinear transformation method to process the frequency domain expression of the fractional-order model obtained in step (5-1) to obtain the transformation form G(z) of the fractional-order model in the z domain.

[0143] Specifically, T s is the sampling period of the system, take T s =1s, the transformed fractional-order circuit model is shown in formula (8).

[0144]

[0145] Among them, a1, a2, a3, a4, and a5 are the parameter identification data of the fractional-order equivalent circuit model, and the mathematical relationship between them and the data to be identified of the fractional-order equivalent circuit model is:

[0146] By utilizing the mathematical relationship between the parameter identification data of the fractional-order equivalent circuit model and the data to be identified of the fractional-order equivalent circuit model, the data to be identified of the parameters of the fractional-order equivalent circuit model can be obtained.

[0147] (5-3) Simplify the transformed form of the fractional-order model obtained in step (5-2) in the z-domain to obtain the recursive equation Y(k) of the fractional-order equivalent circuit model.

[0148] Specifically, the parameter identification using the recursive least squares method based on the minimum forgetting factor requires simplifying the fractional-order equivalent circuit model equation to y(k) = φ(k) T The form of θ(k), where φ(k) = [Y(k-1), Y(k-2), I(k), I(k-1), I(k-2)] is the data vector at time k; θ(k) = [a1, a2, a3, a4, a5] T , is the parameter identification column vector at time k.

[0149] The recursive equation obtained in this step is shown in formula (9).

[0150] Y(k)=a1Y(k-1)+a2Y(k-2)+a3I(k)+a4I(k-1)+a5I(k-2) (9)

[0151] (5-4) Initialize the parameter identification column vector θ(k) and covariance P(k) at time k obtained in step (5-3) to obtain the parameter identification column vector θ(0) and covariance P(0) at time k=0 after initialization.

[0152] Specifically, θ(0)=[0]; P(0) takes a 5-order scalar matrix, and the elements on the diagonal take values ​​of 10e6.

[0153] (5-5) Determine whether k is equal to 60. If so, go to step (5-10), otherwise go to step (5-6).

[0154] (5-6) Use the initialized parameter identification column vector θ(0) and covariance P(0) obtained in step (5-5) to obtain the gain matrix K(k).

[0155] The gain matrix update equation is shown in formula (10):

[0156]

[0157] Where K(k) is the estimated gain at time k; P(k-1) is the covariance at time k-1; λ is the forgetting factor, which ranges from [0.90, 1.00]. Considering the application scenarios of the present invention, 0.955 is preferred.

[0158] (5-7) The parameter identification column vector θ(k) in the recursive equation at time k is updated according to the gain matrix obtained in step (5-6) to obtain the updated parameter identification column vector θ(k) in the recursive equation at time k.

[0159]

[0160] Among them, e(k) is the sampling error at time k.

[0161] (5-8) Obtain the parameter data to be identified of the fractional-order equivalent circuit model based on the parameter identification column vector θ(k) in the updated k-time recursive equation obtained in step (5-7).

[0162] This step specifically uses the θ(k)=[a1,a2,a3,a4,a5] in the above step (5-2). T The mathematical relationship between the parameter data R0, R1, CPE1, R2, and CPE2 to be identified of the fractional-order equivalent circuit model is used to perform mathematical calculations to obtain the parameter data to be identified of the fractional-order equivalent circuit model.

[0163] (5-9) updating the covariance P(k) according to the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) to obtain an updated covariance P(k), and then returning to step (5-5);

[0164] Specifically, the update equation is shown in formula (12):

[0165]

[0166] Wherein, P(k) is the updated covariance at time k; I is a 5-order scalar matrix, and the value range of the elements on the diagonal is [10e5, 10e6], preferably 10e6.

[0167] (5-10) Binary encode the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) and the fractional orders n1 and n2 (n1 and n2 are both 1 at the kth moment) to obtain a gene population expressed in an encoded form.

[0168] (5-11) Obtaining population fitness based on the gene population expressed in coded form obtained in step (5-10).

[0169] Specifically, the population fitness is calculated using the following formula (13):

[0170]

[0171] Where J(i) is the fitness of the population when the population size is i, and i∈[1, N]; N is the population size, preferably N=100; A(i) is the predicted output value U in the fractional-order equivalent circuit model y (y) and the actual measured output value U c The sum of squares of the differences between (y), A max is the optimization target of the genetic algorithm. In the present invention, A max =0.005; U y (j) is the predicted output value at the jth iteration; U c (j) is the actual measured output value at the jth iteration, and j∈[1,M]; M is the number of iterations, preferably M=60;.

[0172] (5-12) The population is subjected to hybridization and mutation processing according to the population fitness obtained in step (5-11) to obtain a processed population.

[0173] Specifically, the hybridization and mutation processing process is shown in formula (14).

[0174]

[0175] Among them, P z (i) is the hybridization probability of the population when the population size is i; K z is the hybridization factor of the population, and its value range is (0.6, 0.9), preferably 0.7; P b (i) is the probability of population mutation when the population size is i; K b is the variation factor, and its value range is [0.01, 0.1], preferably 0.05; A ave is the average fitness in the population.

[0176] (5-13) Decode the population processed in step (5-12) to obtain a decoded population, which is the parameter data to be identified of the fractional-order equivalent circuit model and the fractional orders n1 and n2.

[0177] Specifically, the decoding operation in this step is performed using a binary decoding method, and an inverse transformation is performed to convert the binary-coded genes into a decimal-coded population, namely the parameter data to be identified of the fractional-order equivalent circuit model and the fractional orders n1 and n2.

[0178] (5-14) Determine whether the number of iterations j is greater than 60. If so, proceed to step (5-15). Otherwise, set j=j+1 and return to step (5-11).

[0179] (5-15) Output the parameter data in the fractional-order equivalent circuit, including the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2 of the fractional-order equivalent circuit model, and the fractional orders n1 and n2.

[0180] The advantage of this step (5) is that the recursive least squares method based on the minimum forgetting factor is used to reduce the influence of old data, the changes of the battery system are tracked faster, and the parameter identification is more in line with the system operation characteristics. At the same time, the genetic algorithm is used to identify the fractional order, avoiding the problem that the recursive least squares method based on the minimum forgetting factor cannot identify the fractional order online.

[0181] (6) Input the parameter data obtained by online parameter identification in step (5) into the Fractional Order Model Adaptive Unscented Kalman Filter (FAUKF) for iterative calculation to obtain the SOC at time t t Estimated value and ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t 0,t , the algorithm flow chart is as follows Figure 4 As shown:

[0182] (7) Determine whether the number of iterations in step (6) is greater than 60. If so, proceed to step (8); otherwise, return to step (6).

[0183] (8) Obtain multiple SOC estimation values ​​of the lithium battery for braking energy recovery obtained by 60 iterations and the ohmic internal resistance parameter R of the fractional-order equivalent circuit model obtained by the 60th iteration 0,t .

[0184] Specifically, the SOC of the lithium battery for braking energy recovery is estimated 60 times as a time scale. When outputting, the estimated SOC value of the lithium battery for braking energy recovery obtained by the 60 estimates and the ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model obtained by the latest estimate are output. 0,t .

[0185] The advantage of the above steps (6) to (8) is that the charge state of the lithium battery for braking energy recovery is estimated in real time within a short time scale, which is consistent with the working conditions of the lithium battery for braking energy recovery and ensures the safe and stable operation of the lithium battery for braking energy recovery.

[0186] (9) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t , and obtain the estimated value of the health state (SOH) of the lithium battery used for braking energy recovery through the FAUKF algorithm. The flow chart is as follows Figure 5 shown.

[0187] Specifically, this step characterizes the battery health status by transforming the internal resistance of the battery system, and converts the ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t estimated in step (8) into 0,t The FAUKF algorithm is used as a parameter to iteratively calculate and estimate the health state of the lithium battery used for braking energy recovery, so as to achieve coordinated real-time estimation of the charge state and health state of the lithium battery used for braking energy recovery on micro and macro time scales.

[0188] This step includes the following sub-steps:

[0189] (9-1) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t Obtain the discretized state-space equations and observation equations for estimating the ohmic internal resistance.

[0190] The discretized state space equation for estimating the ohmic internal resistance is shown in formula (20), and the observation equation is shown in formula (21):

[0191] R 0,t+1 =R 0,t +r t (20)

[0192] U 0,t =U oc,t (SOC)-R 0,t IU 1,t -U 2,t +t t (twenty one)

[0193] Among them, R 0,t+1 is the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1; R 0,t is the ohmic internal resistance of the fractional-order equivalent circuit model at time t; r t is the state noise of the state space equation estimating the ohmic internal resistance at time t; t is the observation noise of the observation equation for estimating the ohmic internal resistance at time t.

[0194] (9-2) Initialize the discretized state space equation and observation equation for estimating the ohmic internal resistance obtained in step (9-1).

[0195] Specifically, this step is to initialize the ohmic internal resistance of the fractional-order equivalent circuit model to R0 obtained by parameter identification in step (5); and to initialize the initial ohmic internal resistance covariance of the fractional-order equivalent circuit model to 0.

[0196] (9-3) Perform unscented transformation (UT) on the discrete state space equation and observation equation initialized in step (9-2) to obtain multiple unscented points (sigma points).

[0197] The UT transformation process is shown in the following formula (22):

[0198]

[0199] in, is the mth sigma point at time t, representing the mathematical expectation of the state vector at time t; is the initial predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model; is the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t; m is the dimension of the state variable of the discrete state space equation after initialization in step (9-2); M is the length of the state vector of the discrete state space equation after initialization in step (9-2), and the value is determined by the number of state vectors. In the present invention, M is 3; γ is the scale parameter of the untraceable transformation; α is the divergence factor of the untraceable transformation, which is used to determine the sampling distribution of the sigma point, and the value range is 0.0001-1, preferably 0.005; β is the adjustment parameter of the untraceable transformation, which is used to provide other constraints on the sampling point, and the value is 0.

[0200] (9-4) Apply the multiple sigma points obtained in step (9-3) to the fractional-order equivalent circuit model obtained in step (1) to obtain the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t Variance prediction value P of the state space equation x , the mean weight corresponding to each sigma point and the covariance weights for the discrete state space equations

[0201] Specifically, the relevant calculation formula is shown in formula (23):

[0202]

[0203] Where χ is the weight coefficient, χ = 2; Q t is the noise covariance of the discrete state space equation at time t.

[0204] (9-5) The mean weight corresponding to the sigma point obtained in step (9-4) Covariance Weights for Discrete State-Space Equations Get the observation value of the observation equation at time t+1 and the observed variance predicted value P y .

[0205] Update observations and the observed variance predicted value P y This process is shown in formula (24):

[0206]

[0207] in, is the observation value of the updated observation equation at time t+1; y t is the actual observation value of the observation equation at time t; R t is the observation noise covariance of the observation equation at time t.

[0208] (9-6) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t obtained according to step (9-4) The variance prediction value P of the discrete state space equation x , the mean weight corresponding to the sigma point Covariance Weights for Discrete State-Space Equations And the observation value of the observation equation at time t+1 obtained in step (9-5) and the observed variance predicted value P y , get the input-output cross covariance matrix P xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t .

[0209] The update equation is shown in formula (25):

[0210]

[0211] in, is the transpose of the input-output cross-covariance matrix.

[0212] (9-7) The input-output cross covariance matrix P obtained according to step (9-6) xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t , obtain the noise covariance Q of the discrete state space equation at time t+1 t+1 and the observation noise covariance R t+1 .

[0213] Specifically, the moving window method is used to update the noise, which realizes the adaptive adjustment of the noise covariance of the state space equation and the observation noise covariance of the observation equation, and the process equation is shown in formula (26):

[0214]

[0215] Among them, G k is the adaptive variance of the moving window method; Z is the adaptive moving window.

[0216] (9-8) The noise covariance Q of the discrete state space equation at time t+1 obtained using step (9-7) t+1 and the observation noise covariance R t+1 Get the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1

[0217] Specifically, the noise covariance Q of the discrete state space equation at time t+1 obtained in step (9-7) is t+1 and the observation noise covariance R t+1 Substitute into formulas (23), (24) and (25) to calculate Then calculate The calculation process is shown in formula (27).

[0218]

[0219] (9-9) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1 obtained in step (9-8) Get the estimated SOH value SOH of the lithium battery for braking energy recovery at time t+1 t+1 .

[0220] The calculation formula for the predicted SOH value of the lithium battery for braking energy recovery at time t+1 is shown in formula (28):

[0221]

[0222] Among them, SOH t+1 is the estimated SOH value of the lithium battery for braking energy recovery at time t+1; R N is the maximum available resistance of the lithium battery for braking energy recovery, is the rated ohmic internal resistance of the lithium battery for braking energy recovery; R e It is a healthy termination resistor for lithium batteries used for braking energy recovery. It is usually stipulated that if the capacity of lithium batteries used for braking energy recovery is less than 80% of the rated capacity, the battery must be scrapped. Therefore, R e The size of is the ohmic internal resistance of the lithium battery for braking energy recovery when the capacity of the lithium battery for braking energy recovery is reduced to 80% of the rated capacity.

[0223] The advantage of the above steps (6) to (9) is that the key characteristic parameters SOC and SOH of the lithium battery for brake energy recovery are jointly estimated on a long-short time scale through two FAUKF algorithms, and the estimated result of SOC is used as the input of SOH estimation, which reduces computational redundancy. The SOC and SOH measurement data are updated and corrected with each other, and real-time monitoring can be performed under harsh operating conditions to ensure safe and stable operation of the system.

[0224] (10) Displaying the multiple SOC estimation values ​​of the lithium battery for braking energy recovery obtained in step (8) and the SOH estimation value obtained in step (9) to the user of the electric drive dump truck for mining.

[0225] The advantage of this step (10) is that users of mining electric drive dump trucks can pay real-time attention to the operating status of the lithium battery used for brake energy recovery, promptly detect battery failures and replace aging batteries, thereby ensuring driving safety and improving economic benefits.

[0226] In summary, through the above description of the present invention, the main advantages of the present invention include:

[0227] 1. Establish a second-order RC equivalent circuit model based on the fractional-order model, which can better fit the nonlinear system of the lithium battery used for braking energy recovery in the braking energy recovery system compared to other models;

[0228] 2. The recursive least squares method based on the minimum forgetting factor can reduce the impact of old data, track changes in the battery system more quickly, and make parameter identification more consistent with the system operation characteristics. At the same time, the genetic algorithm is used to identify the fractional order, avoiding the problem that the recursive least squares method based on the minimum forgetting factor cannot identify the fractional order online;

[0229] 3. Using an adaptive unscented Kalman filter algorithm based on a fractional-order model to estimate the state of charge of the regenerative braking battery in real time on a short time scale, ensuring the safe and stable operation of the regenerative braking battery;

[0230] 4. Because the health status of the regenerative braking battery does not change significantly, the battery's state of charge is estimated 60 times over a short time scale, followed by a single health estimate over a longer time scale. This reduces the number of calculations while ensuring estimation accuracy.

[0231] 5. The key characteristic parameters SOC and SOH of the lithium battery used for brake energy recovery are jointly estimated in long-short time scale cycles through two FAUKF algorithms. The estimated SOC result is used as the input for SOH estimation to reduce computational redundancy. The SOC and SOH measurement data are updated and corrected with each other, enabling real-time monitoring under harsh operating conditions to ensure safe and stable operation of the system.

[0232] 6. Users of electric-drive dump trucks for mining can monitor the operating status of the lithium battery used for brake energy recovery in real time, detect battery failures in a timely manner, and replace aging batteries to ensure driving safety and improve economic benefits.

[0233] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm, characterized in that: The following steps are involved: (1) Using SIMULINK in MATLAB software to simulate and model the battery to obtain a fractional-order equivalent circuit model; (2) processing the fractional-order equivalent circuit model obtained in step (1) using Kirchhoff's current law and Kirchhoff's voltage law, respectively, to obtain an electrical principle equation and an output equation of the fractional-order equivalent circuit model; (3) using the Greenwald-Letnikoff formula to process the electrical principle equation and output equation of the fractional-order equivalent circuit model established in step (2) to obtain the state space equation and observation equation defined in fractional order; (4) discretizing the state equation and observation equation defined by the fractional order obtained in step (3) to obtain the discretized state equation and observation equation respectively; (5) using an online parameter identification method to process the discretized state equation and observation equation obtained in step (4) respectively to obtain parameter data in the fractional-order equivalent circuit; the online parameter identification method in step (5) adopts a recursive least square method based on minimum forgetting factor combined with a genetic algorithm; (6) Input the parameter data obtained by online parameter identification in step (5) into the fractional-order adaptive unscented Kalman algorithm FAUKF for iterative calculation to obtain the SOC at time t t Estimated value and ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t 0,t ; (7) Determine whether the number of iterations in step (6) is greater than 60. If so, proceed to step (8); otherwise, return to step (6); (8) Obtain multiple SOC estimation values ​​of the lithium battery obtained by 60 iterations and the ohmic internal resistance parameter R of the fractional-order equivalent circuit model obtained by the 60th iteration 0,t ; (9) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t , and obtain the estimated state of health (SOH) value of the lithium battery through the FAUKF algorithm; Step (9) includes the following sub-steps: (9-1) The ohmic internal resistance parameter R of the battery fractional-order equivalent circuit model at time t obtained according to step (8) 0,t Obtain the discretized state-space equations and observation equations for estimating the ohmic internal resistance: R 0,t+1 =R 0,t +r t The 0,t =U oc,t (SOC)-R 0,t IU 1,t -U 2,t +t t Among them, R 0,t+1 is the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1; R 0,t is the ohmic internal resistance of the fractional-order equivalent circuit model at time t; r t is the state noise of the state space equation estimating the ohmic internal resistance at time t; t is the observation noise of the observation equation for estimating the ohmic internal resistance at time t; (9-2) Initializing the discretized state space equation and observation equation for estimating the ohmic internal resistance obtained in step (9-1); Specifically, this step is to initialize the ohmic internal resistance of the fractional-order equivalent circuit model to R0 obtained by parameter identification in step (5); initialize the initial ohmic internal resistance covariance of the fractional-order equivalent circuit model to 0; (9-3) performing untraceable transformation (UT) on the discrete state space equation and observation equation initialized in step (9-2) to obtain multiple sigma points; (9-4) Apply the multiple sigma points obtained in step (9-3) to the fractional-order equivalent circuit model obtained in step (1) to obtain the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t Variance prediction value P of the state space equation x , the mean weight corresponding to each sigma point and the covariance weights for the discrete state space equations (9-5) The mean weight corresponding to the sigma point obtained in step (9-4) Covariance Weights for Discrete State-Space Equations Get the observation value of the observation equation at time t+1 and the observed variance predicted value P y ; (9-6) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t obtained according to step (9-4) The variance prediction value P of the discrete state space equation x , the mean weight corresponding to the sigma point Covariance Weights for Discrete State-Space Equations And the observation value of the observation equation at time t+1 obtained in step (9-5) and the observed variance predicted value P y , get the input-output cross covariance matrix P xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t ; (9-7) The input-output cross covariance matrix P obtained according to step (9-6) xy , the gain K of the Kalman algorithm at time t t ; The covariance matrix P of the Kalman algorithm at time t t , obtain the noise covariance Q of the discrete state space equation at time t+1 t+1 and the observation noise covariance R t+1 ; (9-8) The noise covariance Q of the discrete state space equation at time t+1 obtained using step (9-7) t+1 and the observation noise covariance R t+1 Get the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1 (9-9) The predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t+1 obtained in step (9-8) Get the estimated SOH value SOH of the lithium battery at time t+1 t+1 ; (10) Displaying the multiple SOC estimation values ​​of the lithium battery obtained in step (8) and the SOH estimation value obtained in step (9) to the user of the mining dump truck.

2. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 1 is characterized in that: The electrical principle equation of the fractional-order equivalent circuit model is: Wherein, U1(t) and U2(t) represent the voltage drops of the two RC loops at time t in the fractional-order equivalent circuit model at the current moment; R1 represents the electrochemical polarization resistance in the fractional-order equivalent circuit model; CPE1 represents the electrochemical polarization capacitance in the fractional-order equivalent circuit model; R2 is the concentration polarization resistance in the fractional-order equivalent circuit model; CPE2 is the concentration polarization capacitance in the fractional-order equivalent circuit model; I t is the operating voltage of the fractional-order equivalent circuit model; n1 and n2 are the fractional orders of the electrochemical polarization capacitance and concentration polarization capacitance in the fractional-order equivalent circuit model, and their values ​​range from 0 to 1; The SOC expression of the state of charge of the lithium battery at the next moment, SOC t is the SOC and SOC of the equivalent circuit model at the current moment t+1 is the SOC of the equivalent circuit model at the next moment; T s is the system sampling period of the equivalent circuit model; Q n is the battery capacity of the equivalent circuit model; The output equation of the fractional-order equivalent circuit model is: U0(t)=U oc (SOC)-R0I(t)-U1(t)-U2(t) Where U0(t) represents the terminal voltage of the fractional-order equivalent circuit model at the current moment, U oc (SOC) represents the open-circuit voltage of the fractional-order equivalent circuit model, R0 represents the ohmic internal resistance of the fractional-order equivalent circuit model, U1(t) and U2(t) are the voltage drops of the two RC loops of the fractional-order equivalent circuit model at the current moment.

3. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 2 is characterized in that: Specifically, the definition formula of fractional derivative is: Among them D n Fractional differential operators representing the state space equations and observation equations; D n f(t) represents the calculus of f(t); n is the order of the fractional-order equivalent circuit model, ranging from [0,1]; n! is the factorial of the order of the fractional-order equivalent circuit model; T s is the system sampling period of the fractional-order equivalent circuit model, and its value range is [0,1]; L is the memory length of the fractional-order equivalent circuit model, and L=60; i is the number of sampling times; is the coefficient of Newton binomial; The state space equation and observation equation defined by the fractional order in step (3) are shown in the following formulas: Among them, [U 1,k U 2,k SOC k ] T is the input of the state-space equation, which represents the voltage drop of loop 1, the voltage drop of loop 2 and the state of charge of the fractional-order equivalent circuit model at the current moment; [U 1,k+1 U 2,k+1 SOC k+1 ] T represents the voltage drop of loop 1, the voltage drop of loop 2, and the state of charge of the fractional-order equivalent circuit model at the next moment; U oc (SOC k ) is the open circuit voltage as a function of the state of charge; I k is the working current of the fractional-order equivalent circuit model at the current moment; ω k is the system noise of the fractional-order equivalent circuit model at time k, which is caused by the uncertainty of the fractional-order equivalent circuit model; k is the measurement noise of the fractional-order equivalent circuit model at time k, which is caused by the measurement error of the fractional-order equivalent circuit model.

4. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 3 is characterized in that: The state equation and observation equation after discretization in step (4) are shown in the following formulas: Among them, x k is the system input of the fractional-order equivalent circuit model at time k, and x k-1 is the system input of the fractional-order equivalent circuit model at time k-1; I k-1 is the operating current of the fractional-order equivalent circuit model at time k-1; ω k-1 is the system noise of the fractional-order equivalent circuit model at time k-1; y k is the system output of the fractional-order equivalent circuit model at time k; C = [-1-10]; υ k is the measurement noise of the fractional-order equivalent circuit model at time k.

5. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 4 is characterized in that: The data to be identified of the fractional-order equivalent circuit include the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2, and fractional orders n1 and n2 of the fractional-order equivalent circuit model.

6. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 5 is characterized in that: Step (5) specifically includes the following sub-steps: (5-1) Perform Laplace transform on the discrete state equation and observation equation obtained in step (4) to obtain the frequency domain expression of the fractional-order model, as shown in formula (7); Among them, Y(k) is the transfer function of the fractional-order equivalent circuit model in the time domain; Y(s) is the transfer function of the fractional-order equivalent circuit model in the frequency domain; s is the independent variable of the transfer function, which is equivalent to the time state in the frequency domain; U oc (s) is the open-circuit voltage of the fractional-order equivalent circuit model in the frequency domain; U o (s) is the terminal voltage of the fractional-order equivalent circuit model in the frequency domain; I(s) is the operating current of the fractional-order equivalent circuit model in the frequency domain; (5-2) Use the bilinear transformation method to process the frequency domain expression of the fractional-order model obtained in step (5-1) to obtain the transformation form G(z) of the fractional-order model in the z domain: Among them, a1, a2, a3, a4, and a5 are the parameter identification data of the fractional-order equivalent circuit model, and the mathematical relationship between them and the data to be identified of the fractional-order equivalent circuit model is: By using the mathematical relationship between the parameter identification data of the fractional-order equivalent circuit model and the data to be identified of the fractional-order equivalent circuit model, the data to be identified of the parameters of the fractional-order equivalent circuit model can be obtained; (5-3) Simplifying the transformed form of the fractional-order model obtained in step (5-2) in the z-domain to obtain the recursive equation Y(k) of the fractional-order equivalent circuit model; Y(k)=a1Y(k-1)+a2Y(k-2)+a3I(k)+a4I(k-1)+a5I(k-2) (5-4) Initializing the parameter identification column vector θ(k) and covariance P(k) at time k obtained in step (5-3) to obtain the parameter identification column vector θ(0) and covariance P(0) at time k=0 after initialization; (5-5) Determine whether k is equal to 60. If so, proceed to step (5-10), otherwise proceed to step (5-6); (5-6) Use the initialized parameter identification column vector θ(0) and covariance P(0) obtained in step (5-5) to obtain the gain matrix K(k): Where K(k) is the estimated gain at time k; P(k-1) is the covariance at time k-1; λ is the forgetting factor, which ranges from [0.90, 1.00]. (5-7) updating the parameter identification column vector θ(k) in the recursive equation at time k according to the gain matrix obtained in step (5-6) to obtain an updated parameter identification column vector θ(k) in the recursive equation at time k; Among them, e(k) is the sampling error at time k; (5-8) obtaining the parameter data to be identified of the fractional-order equivalent circuit model according to the parameter identification column vector θ(k) in the updated k-time recursive equation obtained in step (5-7); (5-9) The covariance P(k) is updated according to the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) to obtain an updated covariance P(k), and then the process returns to step (5-5): Where P(k) is the updated covariance at time k; I is a 5-order scalar matrix with diagonal elements ranging from [10e5, 10e6]; (5-10) Binary encoding the parameter data to be identified of the fractional-order equivalent circuit model obtained in step (5-8) and the fractional orders n1 and n2 to obtain a gene population expressed in a coded form, wherein n1 and n2 are both 1 at the kth moment; (5-11) Obtaining the population fitness based on the gene population expressed in coded form obtained in step (5-10): Where J(i) is the fitness of the population when the population size is i, and i∈[1, N]; N is the population size; A(i) is the predicted output value U in the fractional-order equivalent circuit model y (y) and the actual measured output value U c The sum of the squares of the differences between (y), A max is the optimization goal of the genetic algorithm, A max =0.005; U y (j) is the predicted output value at the jth iteration; U c (j) is the actual measured output value at the jth iteration, and j∈[1,M]; M is the number of iterations; (5-12) According to the population fitness obtained in step (5-11), the population is subjected to hybridization and mutation processing to obtain a processed population: Among them, P z (i) is the hybridization probability of the population when the population size is i; K z is the hybridization factor of the population, and its value range is (0.6, 0.9); P b (i) is the probability of population mutation when the population size is i; K b is the variation factor, and its value range is [0.01, 0.1]; A ave is the average fitness in the population; (5-13) decoding the population processed in step (5-12) to obtain a decoded population, namely, the parameter data to be identified of the fractional-order equivalent circuit model and the fractional orders n1 and n2; (5-14) Determine whether the number of iterations j is greater than 60. If so, proceed to step (5-15). Otherwise, set j = j + 1 and return to step (5-11). (5-15) Output the parameter data in the fractional-order equivalent circuit, including the ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance CPE1, concentration polarization resistance R2, concentration polarization capacitance CPE2 of the fractional-order equivalent circuit model, and the fractional orders n1 and n2.

7. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 6 is characterized in that: The UT transformation process of step (9-3) is: in, is the mth sigma point at time t, representing the mathematical expectation of the state vector at time t; is the initial predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model; is the predicted value of the ohmic internal resistance of the fractional-order equivalent circuit model at time t; m is the dimension of the state variable of the discrete state space equation after initialization in step (9-2); M is the length of the state vector of the discrete state space equation after initialization in step (9-2), and the value is determined by the number of state vectors. In the present invention, M is 3; γ is the scale parameter of the untraceable transformation; α is the divergence factor of the untraceable transformation, which is used to determine the sampling distribution of the sigma point, and the value range is 0.0001-1; β is the adjustment parameter of the untraceable transformation, which is used to provide other constraints on the sampling point, and the value is 0; Step (9-4) uses the following calculation formula: Where χ is the weight coefficient, χ = 2; Q t is the noise covariance of the discrete state space equation at time t.

8. The method for estimating key characteristic parameters of a braking energy recovery battery for a mining dump truck based on a multi-time scale fusion algorithm according to claim 7 is characterized in that: Step (9-5) Update observations and the observed variance predicted value P y The process is as follows: in, is the observation value of the updated observation equation at time t+1; y t is the actual observation value of the observation equation at time t; R t is the observation noise covariance of the observation equation at time t; The update equation of step (9-6) is: in, is the transpose of the input-output cross-covariance matrix; The implementation process of step (9-7) is shown below: Among them, G k is the adaptive variance of the moving window method; Z is the adaptive moving window; The calculation formula for the predicted SOH value of the lithium battery at time t+1 in step (9-9) is as follows: Among them, SOH t+1 is the estimated SOH value of the lithium battery at time t+1; R N is the maximum available resistance of the lithium battery, is the rated ohmic internal resistance of the lithium battery; R e It is the healthy termination resistance of lithium batteries. It is usually stipulated that when the capacity of lithium batteries is less than 80% of the rated capacity, the batteries must be scrapped. Therefore, R e The size of is the ohmic internal resistance of the lithium battery when the capacity of the lithium battery is reduced to 80% of the rated capacity.

Images