Power integration method with energy compensation mechanism for energy state estimation

Abstract

The invention discloses a power integration method with an energy compensation mechanism for energy state estimation, and relates to the technical field of lithium ion batteries. Through analysis of energy state estimation errors and analysis of variable maximum energy and reference maximum energy, a compensation mechanism of an energy state in a power integration method is explored, and a Dropout technology is introduced into the power integration method with energy state compensation. Accumulated errors in the power integration method can be reduced, the energy state estimation precision can be improved, battery aging factors are considered, the applicability of energy compensation is improved, the Dropout technology is integrated during energy compensation, and the robustness and adaptability of the algorithm are improved.

Description

Technical Field

[0001] The invention relates to the technical field of lithium ion batteries, in particular to a power integration method with an energy compensation mechanism for energy state estimation. Background Art

[0002] Lithium-ion batteries are widely used in electric vehicles due to their high energy density, low self-discharge rate, and no memory effect. The battery management system can provide multiple functions such as state estimation, fault diagnosis, and safety warnings by collecting battery voltage, current, and temperature data. Accurate feedback of various battery indicators in electric vehicles is crucial to the driver's driving experience and safety. Among them, accurate feedback of the remaining range is crucial to alleviating the driver's range anxiety. Therefore, accurate estimation of the remaining battery energy is the basis for accurate feedback of the remaining range.

[0003] At present, the methods for estimating the energy state of lithium-ion batteries mainly include coupling calculation method, power integration method, model-based method (such as equivalent circuit method) and data-driven method. Among them, the coupling calculation method first establishes the functional relationship between the energy state and other battery state indicators, and then indirectly estimates the energy state through other battery state estimation indicators. However, due to the estimation errors of other state indicators themselves, these errors will be transmitted to the estimated results of the energy state during the coupling calculation process, which is the shortcoming of this method; the power integration method is a method of calculating energy by integrating power during battery operation. It is simple in form and is also the basis of other energy estimation modeling methods. The power integration method has low computing resource requirements and is suitable for application in embedded devices, but its disadvantage is that as the number of iterations increases, the error will gradually accumulate, eventually leading to a large deviation between the estimated result and the actual energy state; the equivalent circuit method, as a model-based energy state modeling method, still uses the power integration method in the energy state update mechanism, and also fails to effectively solve the problem of accumulated error of the power integration method; with the development of computing resources, data-driven methods have become a novel method for lithium-ion battery energy state estimation. As a nonlinear "black box" system, it can effectively capture the data characteristics of the energy state by establishing a relationship between input features and energy state indicators, but this method is too dependent on the training data set, and its robustness under different working conditions is worthy of further exploration.

[0004] In summary, the power-integrated and data-driven methods currently used in most studies on energy state estimation still have the following problems: the power-integrated modeling method fails to effectively reduce the problem of cumulative errors, resulting in the gradual accumulation of errors in energy state estimation as the number of iterations increases, thus limiting the performance of the power-integrated method. The data-driven method is very effective in capturing the data characteristics of energy state estimation, but it is highly dependent on the quality of the data set, and its robustness needs to be considered when facing unknown working conditions. In addition, the current research on how to introduce the core mechanism of the data-driven method into the relatively simple power-integrated method is still unclear. Summary of the invention

[0005] In order to address the shortcomings of the background technology, the present invention provides a power integration method with an energy compensation mechanism for energy state estimation, which can reduce the cumulative error in the power integration method, improve the accuracy of energy state estimation, consider battery aging factors, enhance the applicability of energy compensation, and incorporate Dropout technology into energy compensation to enhance the robustness and adaptability of the algorithm.

[0006] To achieve the above object, the present invention adopts the following technical solution: a power integration method with energy compensation mechanism for energy state estimation, comprising the following steps:

[0007] Step 1: Analysis of energy state estimation error

[0008] The energy state estimated by the power integration method at the kth iteration is recorded as SOE pi (k), defines the variable maximum energy index E at the kth iteration v (k) is:

[0009]

[0010] in,

[0011] E r,0,k =E r ·[SOE(0)-SOE r (k)] (2)

[0012] The energy state estimation error ΔSOE(k) of the power integration method at the kth iteration is:

[0013] ΔSOE(k)=|SOEpi(k)-SOEr(k)| (4)

[0014] In the formula, E r,0,k represents the cumulative reference energy from the initial state to the kth iteration, SOE(0) represents the initial energy state, Er represents the reference maximum energy index, SOE r (k) represents the reference energy state at the kth iteration;

[0015] Step 2: Analysis of variable maximum energy and reference maximum energy

[0016] The maximum energy error ΔE between the variable maximum energy index and the reference maximum energy index at the kth iteration vr (k) is:

[0017]

[0018] When SOE pi (k) Greater or less than SOE r (k), the maximum energy error ΔE vr (k) Compensation to the cumulative discharge energy E from the initial state to the kth iteration using the power integration method pi,0,k In the case of SOE pi (k) is equal to SOE r (k) No compensation will be made;

[0019] Step 3: Power integration method with energy compensation mechanism

[0020] The ideal form of the power integration method with energy compensation mechanism is:

[0021]

[0022] in,

[0023]

[0024] The energy state compensation considering battery aging is as follows:

[0025]

[0026] In the formula, SOE pic (k) represents the energy state calculated by the power integration method with energy compensation mechanism at the kth iteration, represents the energy conversion coefficient, V(i) and I(i) represent the voltage and current at the i-th iteration, Δt represents the sampling time, and ΔSOE c (i) represents the compensation value of the energy state, ΔE vr (i) represents the maximum energy error at the i-th iteration, ΔSOE ca (i) represents the energy state compensation value considering battery aging at the i-th iteration, Υ=C r / C t , C r Indicates the current maximum cycle capacity, C t Indicates the rated maximum capacity;

[0027] Step 4: Power integration method with energy state compensation under Dropout technology

[0028] Based on step 3, the Dropout technique is used to provide an energy state compensation value ΔSOE for each iteration. ca (i) The compensation value of the random shielding part is finally obtained by the power integration method as follows:

[0029]

[0030] In the formula, Ψ 0～1 represents the shielding probability of energy state compensation.

[0031] Furthermore, the compensation value ΔSOE of the energy state c (i) is a percentage value and at the i-th iteration, it is equal to the discharge energy state Keep the same order of magnitude.

[0032] Compared with the prior art, the beneficial effects of the present invention are: in order to analyze the causes of cumulative errors in the power integration method, the present invention designs a variable maximum energy index, and conducts in-depth analysis based on the index, designs an energy compensation mechanism to reduce the cumulative error in the power integration method, improves the accuracy of energy state estimation, considers battery aging factors, improves the applicability of energy compensation at different battery aging stages, can adapt to different ambient temperatures and charge and discharge rate conditions, incorporates Dropout technology into energy compensation to prevent over-correction of the energy state under unknown operating conditions, improves the robustness and adaptability of the algorithm, and can provide a theoretical basis for accurate energy state estimation, ensure the safe driving of electric vehicles, and improve driving experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flowchart of the method of the present invention;

[0034] Figure 2 It is a principle block diagram of the energy compensation mechanism in the method of the present invention;

[0035] Figure 3 Schematic diagram of applying the Dropout technique in the method of the present invention;

[0036] Figure 4 It is the energy state result diagram of the original power integration method and the method of the present invention in the embodiment;

[0037] Figure 5 It is an energy state estimation error diagram of the original power integration method in the embodiment and the method of the present invention. DETAILED DESCRIPTION

[0038] The technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] like Figure 1 to Figure 3 As shown in the figure, the energy state estimation error is analyzed, the variable maximum energy and the reference maximum energy are analyzed, the compensation mechanism of the energy state in the power integration method is explored, and the Dropout technology is introduced in the power integration method with energy state compensation. A power integration method with energy compensation mechanism for energy state estimation is proposed. Its process is combined with Figure 1 As shown, the following steps are included:

[0040] (1) Analysis of energy state estimation error

[0041] Assume that the energy state estimated by the power integration method at the kth iteration is recorded as SOE pi (k), defines the variable maximum energy index E at the kth iteration v (k) is:

[0042]

[0043] In the formula, E r,0,k represents the cumulative reference energy from the initial state to the kth iteration, and SOE(0) represents the initial energy state.

[0044] The cumulative reference energy from the initial state to the kth iteration can be expressed as:

[0045] E r,0,k =E r ·[SOE(0)-SOE r (k)] (2)

[0046] Where Er represents the reference maximum energy index, SOE r (k) represents the reference energy state at the kth iteration.

[0047] Substituting formula (2) into formula (1), the relationship between the variable maximum energy index and the reference maximum energy index is obtained as follows:

[0048]

[0049] Then the energy state estimation error of the power integration method at the kth iteration is:

[0050] ΔSOE(k)=|SOE pi (k)-SOEr (k)| (4)

[0051] If SOE in formula (4) pi (k) is equal to SOE r (k), E v (k) is equal to E r , which indicates that the variable maximum energy index E v The definition of (k) is reasonable.

[0052] (2) Analysis of variable maximum energy and reference maximum energy

[0053] The maximum energy error between the variable maximum energy index and the reference maximum energy index at the kth iteration is:

[0054]

[0055] The purpose of the present invention is to use the maximum energy error ΔE vr (k) Design compensation methods for energy state estimation to improve SOE pi (k) accuracy, SOE pi (k) and ΔE vr The relationship between (k) is discussed as follows:

[0056] ① When the power integration method calculates the energy state SOE pi (k) greater than the energy state reference value SOE r (k), the cumulative discharge energy E from the initial state to the kth iteration of the power integration method pi,0,k Less than the cumulative reference energy E r,0,k , in E pi,0,k The energy should be increased to reduce E pi,0,k With E r,0,k At the same time, the maximum energy error ΔE vr (k) Satisfy:

[0057]

[0058] Maximum energy error ΔE vr (k) is greater than 0, based on the power integration method, ΔE vr (k) Design compensation methods for energy state estimation to reduce cumulative errors;

[0059] ②When SOE pi (k) Less than SOE r (k), the cumulative discharge energy E from the initial state to the kth iteration of the power integration method pi,0,k Greater than the accumulated reference energy E r,0,k , in E pi,0,k The energy should be reduced to reduce Epi,0,k With E r,0,k At the same time, the maximum energy error ΔE vr (k) Satisfy:

[0060]

[0061] Maximum energy error ΔE vr (k) is less than 0, based on the power integration method, ΔE vr (k) Design compensation methods for energy state estimation to reduce cumulative errors;

[0062] ③When SOE pi (k) is equal to SOE r (k), the cumulative discharge energy E from the initial state to the kth iteration of the power integration method pi,0,k Equal to the accumulated reference energy E r,0,k , E pi,0,k No energy compensation is required. At the same time, the maximum energy error ΔE vr (k) Satisfy:

[0063]

[0064] According to the above analysis, SOE pi (k) and ΔE vr The relationship between (k) is shown in Table 1:

[0065] Table 1 SOE pi (k) and ΔE vr (k) The relationship between

[0066]

[0067] It can be seen that the estimated energy state SOE at the kth iteration calculated by the power integration method is pi (k) and the maximum energy error ΔE vr (k) When SOE pi (k) Greater than SOE r (k), the accumulated discharge energy E pi,0,k Less than reference value E r,0,k , due to ΔE vr (k) is positive, ΔE vr (k) Compensation to E pi,0,k Reduce the cumulative error; when SOE pi (k) Less than SOE r (k), the accumulated discharge energy E pi,0,k Greater than reference value E r,0,k , due to ΔE vr (k) is negative, ΔE vr(k) Compensation to E pi,0,k Reduce the cumulative error; when SOE pi (k) is equal to SOE r (k), the accumulated discharge energy E pi,0,k Equal to reference value E r,0,k , ΔE vr (k) is 0, and no additional energy compensation is required. The above energy state compensation principle is combined with Figure 2 As shown in the figure, the above three types of energy compensation design ideas are given respectively. For known data, the data-driven method is used to obtain ΔE vr (k), the obtained ΔE vr (k) Offline compensation parameter as energy state estimation.

[0068] (3) Power integration method with energy compensation mechanism

[0069] The ideal form of the power integration method with energy compensation mechanism is:

[0070]

[0071] In the formula, SOE pic (k) represents the energy state calculated by the power integration method with energy compensation mechanism at the kth iteration, ζ represents the energy conversion coefficient, V(i) and I(i) represent the voltage and current at the i-th iteration, Δt represents the sampling time, and ΔSOE c (i) represents the compensation value of the energy state.

[0072] Energy state compensation value ΔSOE c (i) is a percentage value and is related to the discharge energy state ζV(i)I(i)Δt / E at the i-th iteration. r Keep the same order of magnitude. According to the maximum energy error in formula (5), the compensation value of the energy state is as follows:

[0073]

[0074] In order to consider the energy change of the battery at different aging stages, the energy state compensation considering battery aging is as follows:

[0075]

[0076] Where, ΔSOE ca (i) represents the energy state compensation value considering battery aging at the i-th iteration, Υ=C r / C t , C r Indicates the current maximum cycle capacity, C t Indicates the rated maximum capacity, ΔE vr(i) represents the maximum energy error at the i-th iteration.

[0077] (4) Power integration method with energy state compensation under Dropout technology

[0078] The energy state compensation method considering battery aging is designed as shown in formula (11), but the maximum energy error ΔE obtained by training the data-driven method is vr (k) may not be applicable to different temperatures, charge and discharge rates, or charge and discharge strategies. Taking temperature as an example, due to the differences in the discharge characteristics of lithium-ion batteries at different temperatures, the maximum energy error measured at 25°C may not be directly used to compensate for the energy state at 5°C. Assuming that the energy state compensation value obtained at 25°C in formula (11) is directly applied to the energy state estimation at 5°C, the problem of "overcorrection" may occur, that is, the use of the energy state compensation mechanism will cause the estimation result to deviate from the reference value, as shown below:

[0079]

[0080] In the formula, K represents the total number of iterations.

[0081] The "overcorrection" problem in formula (12) is due to the difference in battery discharge characteristics under different working conditions, which leads to correction deviation in each iteration of the power integration method. The core idea of ​​the Dropout technology is to randomly shield some nodes. The present invention uses the Dropout technology to solve the "overcorrection" problem in formula (12) and improve the robustness of the energy state compensation mechanism. The Dropout technology in the power integration method with energy state compensation is combined with Figure 3 In the power integration method based on energy state compensation, an energy state compensation value ΔSOE is provided at each iteration. ca (i) However, a part of the compensation value is randomly shielded and no energy compensation is performed at the shielded location, which can solve the "overcorrection" problem in energy state compensation.

[0082] Combining the energy state compensation of formula (9) and formula (11) and the Dropout technology, the final power integration method is as follows:

[0083]

[0084] In the formula, Ψ 0～1 represents the shielding probability of energy state compensation.

[0085] In addition to being applicable to the power integration method, the present invention is also applicable to other energy state modeling methods based on the power integration method, and the design ideas of the energy compensation mechanisms therein are consistent.

[0086] Example

[0087] The verification was conducted based on a nickel cobalt manganese oxide lithium ion battery with a rated capacity of 2000 mAh, and dynamic test experimental data were obtained under FUDS working conditions. Based on the power integration method with energy compensation of the present invention, the energy state estimation result was obtained by running on the PyTorch platform.

[0088] Combination Figure 4 As shown, there is a comparison between the estimation results of the original power integration method and the power integration method with energy compensation of the present invention and the reference value. It can be seen that the estimation result of the original power integration method is obviously far away from the reference value, while the estimation result of the power integration method with energy compensation of the present invention is closer to the reference value, and the energy state estimation accuracy using the energy compensation mechanism is higher.

[0089] Combination Figure 5 As shown, it is a comparison of the energy state estimation error of the original power integration method and the power integration method with energy compensation of the present invention. It can be seen that the power integration method with energy compensation of the present invention can significantly reduce the cumulative error and is more reasonable and effective.

[0090] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other forms of assembly without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0091] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A power integration method with an energy compensation mechanism for energy state estimation, characterized in that: The following steps are involved: Step 1: Analysis of energy state estimation error The energy state estimated by the power integration method at the kth iteration is recorded as SOE pi (k), defines the variable maximum energy index E at the kth iteration v (k) is: in, E r,0,k =E r ·[SOE(0)-SOE r (k)] (2) The energy state estimation error ΔSOE(k) of the power integration method at the kth iteration is: ΔSOE(k)=|SOE pi (k)-SOE r (k)| (4) In the formula, E r,0,k represents the cumulative reference energy from the initial state to the kth iteration, SOE(0) represents the initial energy state, Er represents the reference maximum energy index, SOE r (k) represents the reference energy state at the kth iteration; Step 2: Analysis of variable maximum energy and reference maximum energy The maximum energy error ΔE between the variable maximum energy index and the reference maximum energy index at the kth iteration vr (k) is: When SOE pi (k) Greater or less than SOE r (k), the maximum energy error ΔE vr (k) Compensation to the cumulative discharge energy E from the initial state to the kth iteration using the power integration method pi,0,k In the case of SOE pi (k) is equal to SOE r (k) No compensation will be made; Step 3: Power integration method with energy compensation mechanism The ideal form of the power integration method with energy compensation mechanism is: in, The energy state compensation considering battery aging is as follows: In the formula, SOE pic (k) represents the energy state calculated by the power integration method with energy compensation mechanism at the kth iteration, represents the energy conversion coefficient, V(i) and I(i) represent the voltage and current at the i-th iteration, Δt represents the sampling time, and ΔSOE c (i) represents the compensation value of the energy state, ΔE vr (i) represents the maximum energy error at the i-th iteration, ΔSOE ca (i) represents the energy state compensation value considering battery aging at the i-th iteration, Υ=C r / C t , C r Indicates the current maximum cycle capacity, C t Indicates the rated maximum capacity; Step 4: Power integration method with energy state compensation under Dropout technology Based on step 3, the Dropout technique is used to provide an energy state compensation value ΔSOE for each iteration. ca (i) The compensation value of the random shielding part is finally obtained by the power integration method as follows: In the formula, Ψ 0～1 represents the shielding probability of energy state compensation.

2. A power integration method with energy compensation mechanism for energy state estimation according to claim 1, characterized in that: The compensation value of the energy state ΔSOE c (i) is a percentage value and at the i-th iteration, it is equal to the discharge energy state Keep the same order of magnitude.

Images