Battery multi-dimensional health index joint estimation method applied to electric vertical take-off and landing aircraft

Abstract

The invention discloses a battery multi-dimensional health index joint estimation method applied to an electric vertical take-off and landing aircraft. The method comprises the steps that model input and output parameter matrixes are acquired offline; normalizing the input and output data; training the estimation model by adopting a multi-output least square support vector regression algorithm; recording battery current and voltage data in a take-off stage; normalizing the recorded data and substituting the data into the estimation model; and performing reverse normalization processing to output health indexes. The method provided by the invention has the following three advantages: (1) the battery energy and the power decline are comprehensively considered, and the health state of the airborne battery system can be reflected more comprehensively and accurately; (2) battery current and voltage data sets at the take-off stage are used as estimation model input, so that battery aging information can be comprehensively mined, and the accuracy and generalization of the model are improved; and (3) the multi-output least square support vector regression algorithm used in the invention improves the estimation precision while ensuring the model training efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of battery technology and new energy aircraft, and in particular relates to a method for jointly estimating multi-dimensional health indicators of batteries applied to electric vertical take-off and landing aircraft. Background Art

[0002] Electric Vertical Take-Off and Landing aircraft (eVTOL) with electric propulsion system has received extensive attention and great development due to its low noise and zero emission characteristics, and has gradually become an important trend in the development of low-altitude aircraft. Among them, the power battery system, as the main or only power source of eVTOL, directly determines the feasibility and safety of the aircraft. Lithium-ion batteries are gradually being used in light and small aircraft due to their high specific energy / specific power, low self-discharge rate and good cycle stability. Compared with the fields of electric vehicles and grid energy storage, the current research on power battery systems used in aircraft is still in its infancy.

[0003] As the performance of lithium-ion battery systems declines, their safe operating range gradually changes. In order to quantitatively describe the energy decay and power attenuation characteristics of the battery, the capacity and internal resistance parameters of the battery are generally selected as health indicators. Among them, the reduction of battery capacity will directly shorten the range of the aircraft; the increase of internal resistance will compress the battery's available state of charge range, which will not only shorten the range of the aircraft, but also reduce the safety and reliability of its landing phase. Therefore, it is necessary to comprehensively and accurately perceive the capacity and internal resistance parameters of the onboard battery in combination with the operating characteristics of the eVTOL to ensure efficient, reliable and safe flight. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a joint estimation method for multi-dimensional health indicators of batteries applied to electric vertical take-off and landing aircraft. The estimation method is based on the battery current and voltage data during the take-off phase of the electric vertical take-off and landing aircraft, and adopts a multi-output least squares support vector regression algorithm to comprehensively and accurately estimate the three-dimensional health indicators of battery capacity, ohmic internal resistance and polarization internal resistance.

[0005] The present invention is achieved through the following technical solutions.

[0006] The present invention provides a method for jointly estimating multi-dimensional health indicators of a battery applied to an electric vertical take-off and landing aircraft, the method comprising the following steps:

[0007] S1, obtaining the input parameter matrix and output parameter matrix of the battery health index estimation model according to the aging test data of the power battery for the electric vertical take-off and landing aircraft;

[0008] S2, normalizing the input parameter set and the output parameter set data obtained in step S1 to serve as a training set for a battery multi-dimensional health indicator estimation model;

[0009] S3, based on the training set obtained in step S2, using a multi-output least squares support vector regression algorithm, constructing and storing a battery multi-dimensional health indicator joint estimation model based on the take-off phase data of the electric vertical take-off and landing aircraft;

[0010] S4, during the take-off phase of the electric vertical take-off and landing aircraft, recording and storing the battery time-current sequence and time-voltage sequence;

[0011] S5, when the take-off phase is over, normalizing the current and voltage data recorded in step S4 and forming a normalized input parameter set;

[0012] S6, substituting the normalized input parameter set obtained in step S5 into the battery multi-dimensional health index joint estimation model obtained in step S3 to obtain a model output parameter set;

[0013] S7, denormalize the model output parameter set obtained in step S6 to obtain a multi-dimensional battery health index.

[0014] In the step S1, the battery current and voltage data of the electric vertical take-off and landing aircraft in the take-off phase in the data set are screened and sorted into the input parameter matrix of the battery health index estimation model, and the input parameter variable dimension is 2t phase / T s , where t phase is the duration of the take-off phase, T s is the sampling period.

[0015] In the step S1, the battery capacity, the ohmic internal resistance, and the sum of the ohmic internal resistance and the polarization internal resistance are used as the output parameter matrix of the battery health index estimation model, wherein the battery capacity is obtained by using the ampere-hour integration method for the time-current series data of the constant current discharge stage in the reference performance test; the calculation expression of the battery ohmic internal resistance is:

[0016]

[0017] Among them, R o is the battery ohmic internal resistance, V t,0 and V t,1 are the battery terminal voltage values ​​at the initial moment of step current excitation and 1 second later, I L,0 and I L,1 are the current values ​​at the initial moment of step current excitation and 1 second later respectively. The calculation expression of the sum of the battery ohmic internal resistance and polarization internal resistance is:

[0018]

[0019] Among them, R p is the battery polarization internal resistance, V t,30 and I L,30 They are the battery terminal voltage and current values ​​after 30 seconds of step current excitation.

[0020] In step S2, the input parameter matrix and the output parameter matrix are combined to form a training set. The number of samples in the training set is the number of aging cycles of the batteries in the data set. The expression for normalizing the training set parameters is:

[0021]

[0022] Among them, x i is the training set parameter, x i,norm is the normalized training set parameter, x i,min and x i,max are the minimum and maximum values ​​of the same type of data set in the training set, respectively.

[0023] In step S3, the training steps of the battery multi-dimensional health index estimation model based on the multi-output least squares support vector regression algorithm are:

[0024] S3.1 set the hyperparameter set;

[0025] S3.2 calculates and obtains the weight coefficient matrix W and the deviation vector b;

[0026] S3.3 obtains the battery multi-dimensional health index estimation model and calculates the training error. The battery multi-dimensional health index estimation model expression is:

[0027]

[0028] Where f(·) is the output regression function of the battery multi-dimensional health index estimation model, X is the input parameter matrix, is a nonlinear mapping function; the training error expression is:

[0029]

[0030] Among them, RMSE is the root mean square error, y i,ref and i,est are the reference value and estimated value of the battery health index, respectively, and l is the number of aging cycles of the battery in the dataset;

[0031] S3.4 determines whether the training model accuracy meets the requirements; if it does, execute step S3.5; otherwise, execute step S3.1;

[0032] S3.5 saves the battery multi-dimensional health indicator estimation model.

[0033] In step S7, the denormalization expression of the model output parameter is:

[0034] y i,anti-norm =y i (y i,max -y i,min )+y i,min

[0035] Among them, y i,anti-norm is the model output parameter set after denormalization, y i is the model output parameter set, y i,min and i,max They are the minimum and maximum values ​​of the model output parameter set corresponding to the same type of data set.

[0036] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0037] 1. The present invention uses battery capacity, ohmic internal resistance, and the sum of ohmic internal resistance and polarization internal resistance as the output parameter matrix of the battery health index estimation model, comprehensively considering the battery energy decay and power attenuation characteristics, and can more comprehensively and accurately reflect the health status of the battery system for electric vertical take-off and landing aircraft;

[0038] 2. Different from the battery health status estimation method based on voltage feature points during flight, the present invention uses the battery current and voltage data set of the electric vertical take-off and landing aircraft during the take-off phase as the input parameter matrix of the battery health index estimation model, which can more comprehensively and accurately mine the battery aging information and improve the accuracy and generalization of the trained model;

[0039] 3. The battery multi-dimensional health index joint estimation method based on the multi-output least squares support vector regression algorithm constructed by the present invention takes into account the coupling relationship between the output variables, while ensuring the efficiency of model training, and improving the overall accuracy of the model output. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 The figure is a flow chart of the battery health index joint estimation method of the present invention.

[0041] Figure 2 The present invention is a training flow chart of a battery multi-dimensional health index estimation model based on a multi-output least squares support vector regression algorithm.

[0042] Figure 3 The estimated value of the battery capacity estimated by the present invention is compared with the actual measured value.

[0043] Figure 4 The estimated value of the battery internal resistance estimated by the present invention is compared with the actual measured value. DETAILED DESCRIPTION

[0044] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0045] The embodiment of the present invention provides a method for jointly estimating multi-dimensional health indicators of batteries applied to electric vertical take-off and landing aircraft, so as to achieve efficient and safe management of airborne power batteries.

[0046] A joint estimation method for multi-dimensional health indicators of batteries applied to electric vertical take-off and landing aircraft is as follows: Figure 1 As shown, the method is mainly divided into two parts; the first part is the offline training part, and the second part is the online state estimation part; the two parts will be further explained below.

[0047] The offline training part includes the following steps:

[0048] 1) Based on the aging test data of the power battery for electric vertical take-off and landing aircraft, the input parameter matrix and output parameter matrix of the battery health index estimation model are obtained;

[0049] Among them, the battery capacity, ohmic internal resistance, and the sum of ohmic internal resistance and polarization internal resistance are used as the output parameter matrix of the battery health index estimation model, wherein the battery capacity is obtained by using the ampere-hour integration method for the time-current series data of the constant current discharge stage in the reference performance test; the calculation expression of the battery ohmic internal resistance is:

[0050]

[0051] Among them, R o is the battery ohmic internal resistance, V t ,0 and V t , 1 are the battery terminal voltage values ​​at the initial moment of step current excitation and 1 second later, I L ,0 and I L , 1 are the current values ​​at the initial moment of step current excitation and 1 second later respectively. The calculation expression of the sum of the battery ohmic internal resistance and polarization internal resistance is:

[0052]

[0053] Among them, R p is the battery polarization internal resistance, V t,30 and I L,30 They are the battery terminal voltage and current values ​​after 30 seconds of step current excitation.

[0054] 2) normalizing the input parameter set and output parameter set data obtained in step 1) as a training set for the battery multi-dimensional health indicator estimation model;

[0055] Among them, the input parameter matrix and the output parameter matrix are combined to form a training set. The number of training set samples is the number of aging cycles of the batteries in the data set. The expression for normalizing the training set parameters is:

[0056]

[0057] Among them, x i is the training set parameter, x i,norm is the normalized training set parameter, x i,min and x i,max are the minimum and maximum values ​​of the same type of data set in the training set, respectively.

[0058] 3) Based on the training set obtained in step 2), a multi-output least squares support vector regression algorithm is used to construct and store a battery multi-dimensional health indicator joint estimation model based on the take-off phase data of the electric vertical take-off and landing aircraft;

[0059] Among them, the training steps of the battery multi-dimensional health index estimation model based on the multi-output least squares support vector regression algorithm are as follows: Figure 2 As shown, specifically:

[0060] 3.1) Set the hyperparameter set;

[0061] 3.2) Calculate and obtain the weight coefficient matrix W and the deviation vector b;

[0062] 3.3) Obtain the battery multi-dimensional health indicator estimation model and calculate the training error.

[0063] The type expression is:

[0064]

[0065] Where f(·) is the output regression function of the battery multi-dimensional health index estimation model, X is the input parameter matrix, is a nonlinear mapping function; the training error expression is:

[0066]

[0067] Among them, RMSE is the root mean square error, y i,ref and i,est are the reference value and estimated value of the battery health index, respectively, and l is the number of aging cycles of the battery in the dataset;

[0068] 3.4) Determine whether the training model accuracy meets the requirements; if it does, proceed to step 3.5); otherwise, proceed to step 3.1);

[0069] 3.5) Save the battery multi-dimensional health indicator estimation model.

[0070] The online status estimation part includes the following steps:

[0071] 1) During the take-off phase of the electric vertical take-off and landing aircraft, record and store the battery time-current sequence and time-voltage sequence;

[0072] 2) After the take-off phase is over, the current and voltage data recorded in step 1) are normalized and form a normalized input parameter set;

[0073] 3) Substituting the normalized input parameter set obtained in step 2) into the battery multi-dimensional health index joint estimation model obtained in step 3) of the offline training part to obtain the model output parameter set;

[0074] 4) Denormalize the model output parameter set obtained in step 3) to obtain the multi-dimensional health index of the battery.

[0075] Among them, the denormalization expression of the model output parameters is:

[0076] y i,anti-norm =y i (y i,max -y i,min )+y i,min

[0077] Among them, y i,anti-norm is the model output parameter set after denormalization, y i is the model output parameter set, y i,min and i,max They are the minimum and maximum values ​​of the model output parameter set corresponding to the same type of data set.

[0078] In this embodiment, the implementation object is a ternary lithium-ion battery with a nominal capacity of 3.0Ah, but it is not limited to this in practical applications. The estimated values ​​of battery capacity and internal resistance obtained by the present invention are compared with the actual measured values ​​as shown in the following figure. Figure 3 and Figure 4 It can be found that the estimated values ​​of battery capacity and internal resistance can track the reference values ​​well and capture the corresponding nonlinear evolution trend. Therefore, this method can well achieve a comprehensive and accurate characterization of the battery aging state.

[0079] In summary, the present invention provides a method for jointly estimating multi-dimensional health indicators of batteries applied to electric vertical take-off and landing aircraft, including: obtaining an input parameter matrix and an output parameter matrix of a battery health indicator estimation model based on aging test data of power batteries for electric vertical take-off and landing aircraft; normalizing the input parameter set and the output parameter set data as a training set of the battery multi-dimensional health indicator estimation model; using a multi-output least squares support vector regression algorithm to construct and store a battery multi-dimensional health indicator joint estimation model based on the take-off phase data of the electric vertical take-off and landing aircraft; during the take-off phase of the electric vertical take-off and landing aircraft, recording and storing the battery time-current series and time-voltage series; after the take-off phase, normalizing the recorded current and voltage data to form a normalized input parameter set; substituting the normalized input parameter set into the battery multi-dimensional health indicator joint estimation model to obtain a model output parameter set; and denormalizing the model output parameter set to obtain a battery multi-dimensional health indicator. The estimation method proposed in the present invention has the following three advantages: (1) The present invention comprehensively considers the battery energy decay and power attenuation characteristics, and can more comprehensively and accurately reflect the health status of the battery system for electric vertical take-off and landing aircraft; (2) Compared with the battery health status estimation method based on the voltage characteristic points during the flight, the battery health index estimation method based on the battery current and voltage data set in the take-off phase provided by the present invention can more comprehensively and accurately mine the battery aging information, thereby improving the accuracy and generalization of the trained model; (3) The battery multi-dimensional health index joint estimation method constructed by the present invention takes into account the coupling relationship between the output variables, thereby improving the overall accuracy of the model output while ensuring the efficiency of model training.

[0080] Those skilled in the art can understand that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0081] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0082] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for jointly estimating multi-dimensional health indicators of batteries for electric vertical take-off and landing aircraft, characterized in that: The method comprises the following steps: S1, obtaining the input parameter matrix and output parameter matrix of the battery health index estimation model according to the aging test data of the power battery for the electric vertical take-off and landing aircraft; S2, normalizing the input parameter set and the output parameter set data obtained in step S1 to serve as a training set for a battery multi-dimensional health indicator estimation model; S3, based on the training set obtained in step S2, using a multi-output least squares support vector regression algorithm, constructing and storing a battery multi-dimensional health indicator joint estimation model based on the take-off phase data of the electric vertical take-off and landing aircraft; S4, during the take-off phase of the electric vertical take-off and landing aircraft, recording and storing the battery time-current sequence and time-voltage sequence; S5, when the take-off phase is over, normalizing the current and voltage data recorded in step S4 and forming a normalized input parameter set; S6, substituting the normalized input parameter set obtained in step S5 into the battery multi-dimensional health index joint estimation model obtained in step S3 to obtain a model output parameter set; S7, denormalize the model output parameter set obtained in step S6 to obtain a multi-dimensional battery health index.

2. A method according to claim 1, characterized in that In the step S1, the battery current and voltage data of the electric vertical take-off and landing aircraft in the take-off phase in the data set are screened and sorted into the input parameter matrix of the battery health index estimation model, and the input parameter variable dimension is 2t phase / T s , where t phase is the duration of the take-off phase, T s is the sampling period.

3. A method according to claim 1, characterized in that In the step S1, the battery capacity, the ohmic internal resistance, and the sum of the ohmic internal resistance and the polarization internal resistance are used as the output parameter matrix of the battery health index estimation model, wherein the battery capacity is obtained by using the ampere-hour integration method for the time-current series data of the constant current discharge stage in the reference performance test; the calculation expression of the battery ohmic internal resistance is: Among them, R o is the battery ohmic internal resistance, V t,0 and V t,1 are the battery terminal voltage values ​​at the initial moment of step current excitation and 1 second later, I L,0 and I L,1 are the current values ​​at the initial moment of step current excitation and 1 second later respectively. The calculation expression of the sum of the battery ohmic internal resistance and polarization internal resistance is: Among them, R p is the battery polarization internal resistance, V t,30 and I L,30 They are the battery terminal voltage and current values ​​after 30 seconds of step current excitation.

4. A method according to claim 1, characterized in that In step S2, the input parameter matrix and the output parameter matrix are combined to form a training set. The number of samples in the training set is the number of aging cycles of the batteries in the data set. The expression for normalizing the training set parameters is: Among them, x i is the training set parameter, x i,norm is the normalized training set parameter, x i,min and x i,max are the minimum and maximum values ​​of the same type of data set in the training set, respectively.

5. A method according to claim 1, characterized in that: In step S3, the training steps of the battery multi-dimensional health index estimation model based on the multi-output least squares support vector regression algorithm are: S3.1 set the hyperparameter set; S3.2 calculates and obtains the weight coefficient matrix W and the deviation vector b; S3.3 obtains the battery multi-dimensional health index estimation model and calculates the training error. The battery multi-dimensional health index estimation model expression is: Where f(·) is the output regression function of the battery multi-dimensional health index estimation model, X is the input parameter matrix, is a nonlinear mapping function; the training error expression is: Among them, RMSE is the root mean square error, y i,ref and i,est are the reference value and estimated value of the battery health index, respectively, and l is the number of aging cycles of the battery in the dataset; S3.4 determines whether the training model accuracy meets the requirements; if it does, execute step S3.5; otherwise, execute step S3.1; S3.5 saves the battery multi-dimensional health indicator estimation model.

6. A method according to claim 1, characterized in that: In step S7, the denormalization expression of the model output parameter is: and i,anti-norm =and i (and i,max -and i,min )+y i,min Among them, y i,anti-norm is the model output parameter set after denormalization, y i is the model output parameter set, y i,min and i,max They are the minimum and maximum values ​​of the model output parameter set corresponding to the same type of data set.

Images