A power battery temperature estimation method based on polyhedral filtering under double disturbance

Abstract

The present invention discloses a method for estimating the temperature of a power battery based on polyhedral filtering under dual disturbance, and belongs to the technical field of power battery state estimation. The method includes establishing an electric thermal model of a power battery mixed with unknown but bounded noise and Gaussian noise, designing a polyhedral update strategy based on the polyhedral space model and the probability distribution of known noise, constructing a state estimator, and obtaining the optimal weights of the noise and the unknown but bounded noise that meet a certain probability distribution based on the minimum mean square error and the minimum polyhedral space, rather than taking the same weights, thereby improving the accuracy of the estimated interval containing the true value of the state. Compared with the existing state estimation method combining random and set membership mixed noise dual filtering, the state estimation accuracy of the method of the present application is higher, and since the result interval is smaller, the specific value of the estimate can be further determined.

Description

Technical Field

[0001] The invention relates to a power battery temperature estimation method based on polyhedral filtering under double disturbance, belonging to the technical field of power battery state estimation. Background Art

[0002] As a green and clean chemical energy source, power batteries have a broad research background in scientific research and engineering practice. The widespread application of power batteries places high demands on battery performance. Safety and reliability are the most critical issues in battery energy storage systems. In order to ensure the safe operation of the entire energy storage system, it is necessary to perform real-time and accurate state estimation of the battery's operating status.

[0003] As we all know, there is a lot of noise in the battery operation state, so the estimation of the battery operation state needs to consider the noise problem. Traditional state estimation methods usually assume that the noise existing in the operation process of the power battery system meets a certain probability distribution, and use random noise methods such as Kalman filtering to achieve state estimation; but the working environment of the power battery is complex, and the probability distribution of some noise in the system noise is difficult to obtain completely, or it does not meet the certain probability distribution at all, which leads to inaccurate state estimation results. In order to solve this problem, some scholars have proposed a set membership estimation method, which only requires the system noise to be bounded and does not need to know the statistical characteristics such as noise distribution. When representing the state estimation area, geometric bodies such as ellipsoids, intervals, and polytopes can be used for representation, but the set membership estimation obtains the feasible set of state variables rather than specific estimated values. At the same time, because this method does not specifically consider the noise that meets a certain probability distribution, the feasible set of the state estimation results obtained by it has room for further contraction.

[0004] As can be seen from the above, the existing state estimation methods only calculate a certain type of uncertain noise. For dual disturbance systems with two types of noise, using only random noise methods or set membership estimation methods alone is not conducive to the estimation results. In order to fully utilize the advantages of the two methods and ensure reliable estimation results, the research on dual filtering technology combining two types of noise has also received attention in the field of state estimation. However, in some current state estimation studies on dual filtering, the same weights are selected to process various random and unknown but bounded mixed noises during the estimation process. However, the various random noises and unknown but bounded noises in the complex actual operating environment cannot be the same, resulting in the estimation results obtained by such state estimation methods being not accurate enough. Summary of the invention

[0005] In order to solve the problem that the current battery state estimation accuracy is low under the dual disturbance of random and unknown but bounded mixed noise, the present invention provides a power battery temperature estimation method based on polyhedral filtering under dual disturbance, the method comprising:

[0006] Step 1: Establish a power battery electrothermal model with a mixture of unknown but bounded noise and Gaussian noise;

[0007] Step 2: Based on the optimal estimate of the temperature at time k Determine the predicted value of the temperature at time k+1 Consider the corresponding prediction errors for unknown but bounded noise and Gaussian noise respectively;

[0008] Step 3: Based on the predicted value of temperature at time k+1 Determine the best estimate of the temperature at time k+1 Consider the corresponding estimation errors for unknown but bounded noise and Gaussian noise respectively and determine the optimal weight coefficient at time k+1 to achieve a trade-off between the two noises;

[0009] Step 4: Consider the influence of unknown but bounded noise and calculate the optimal estimate of the temperature at time k+1 The temperature estimation interval at time k+1 is obtained to complete the temperature estimation of the power battery under the operating state.

[0010] Optionally, the power battery electrothermal model of the unknown but bounded noise mixed with Gaussian noise established in Step 1 is:

[0011]

[0012] Among them, x k =[x 1,k ,x 2,k ] T =[T c,k ,T s,k ] T is the temperature variable, T c,k and T s,k They represent the battery core temperature and surface temperature at time k, respectively, and u k =[Q gen,k ,T e,k ] T is the input matrix at time k, Q gen Indicates the heat generation power of the battery core; T e,k represents the ambient temperature at time k, w k ∈<0,W> represents the unknown but bounded disturbance noise at time k, represents the disturbance noise that follows Gaussian distribution at time k; v k ∈<0,V> represents the unknown but bounded measurement noise at time k, represents the measurement noise that follows Gaussian distribution at time k; A represents the state space matrix, B represents the input matrix, and C represents the output matrix; y kis the output variable, i.e. the temperature measurement value of the power battery under the operating state.

[0013] Optionally, the Step 2 includes:

[0014] If k=0, the optimal estimated value at the initial moment is set according to the core temperature and surface temperature of the power battery at the initial operation moment, and then the predicted value of the temperature at the next moment is determined according to the optimal estimated value of the temperature at the initial moment;

[0015] If k≠0, the optimal estimated value at time k is set according to the core temperature and surface temperature of the power battery at time k, and then the optimal estimated value of the temperature at time k is set according to the optimal estimated value of the temperature at time k. Determine the predicted value of the temperature at time k+1

[0016]

[0017]

[0018]

[0019] in, represents the predicted point estimate of the temperature at time k+1, Indicates that k+1 time follows Gaussian distribution Partial prediction error, covariance matrix P k+1|k for:

[0020] P k+1|k =AP k A T +Q (5)

[0021] Among them, P k is the covariance matrix of the estimation error that follows the Gaussian distribution at time k; is the covariance matrix of the disturbance noise that obeys Gaussian distribution;

[0022] represents the prediction error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1|k > indicates that the generator matrix H k+1|k for:

[0023] H k+1|k =[AH k W] (6)

[0024] Among them, H k is the generator matrix of the unknown but bounded part of the estimation error at time k; W is the generator matrix of the unknown but bounded disturbance noise.

[0025] Optionally, the Step 3 includes:

[0026] The optimal estimate of the state at time k+1 is calculated according to the following formula (7):

[0027]

[0028]

[0029]

[0030] Where I is the identity matrix, y k+1 is the output of the system at time k+1, represents the point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial estimation error, covariance matrix P k+1 for:

[0031]

[0032] C is the output matrix; R is the covariance matrix of the measurement noise that obeys the Gaussian distribution;

[0033] represents the estimated error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1 > indicates that the generator matrix H k+1 for:

[0034] H k+1 =[(IK k+1 C)H k+1|k K k+1 V] (11)

[0035] V is the generator matrix of the unknown but bounded measurement noise;

[0036] K k+1 is the optimal gain at time k+1. According to (10) and (11), the optimal criterion is:

[0037]

[0038] Among them, tr(P k+1 ) indicates P k+1 traces, Indicates H k+1 The transpose of k+1 Represents the optimal weight coefficient at time k+1:

[0039]

[0040] Solve the optimal criterion shown in equation (12) to obtain the optimal gain K k+1 :

[0041]

[0042] Among them, M k+1 =[H k+1|k 0],

[0043] The optimal gain K k+1 Substitute into formula (8) to obtain the point estimate of the state at time k+1 under the optimal gain;

[0044] The optimal gain K k+1 Substituting into formula (10) and (11) respectively, we can obtain the Gaussian distribution at time k+1 under the optimal gain: The estimation error of the part, and the estimation error of the unknown but bounded part at time k+1 under the optimal gain;

[0045] Then, the optimal estimate is calculated according to formula (7):

[0046] Optionally, the Step 4 includes:

[0047] Considering the influence of unknown but bounded noise, setting the tolerance interval, according to the optimal estimate of the temperature at time k+1 Determine the temperature estimation interval at time k+1 and complete the temperature estimation of the power battery under the operating state

[0048]

[0049] in, and They represent the upper and lower bounds of the Gaussian state estimation error confidence interval, and represent the upper and lower bounds of the unknown but bounded state estimation error interval, respectively.

[0050] Optionally, the Step 1 includes:

[0051] According to the electrochemical mechanism of the power battery, an equivalent circuit of the power battery is established. According to the equivalent circuit of the power battery, the following equation is obtained:

[0052]

[0053] Among them, R1 and C p1 R1 and C are the electrochemical polarization internal resistance and capacitance of the power battery, respectively. p1 The voltage across the two terminals after they are connected in parallel is U1; R2 and C p2 R2 and C are the concentration polarization resistance and capacitance of the power battery respectively. p2The voltage across the two terminals after they are connected in parallel is U2; I is the discharge current;

[0054] According to the thermal model of the power battery, an electrothermal model of the power battery with unknown but bounded noise mixed with Gaussian noise is established;

[0055] According to the heat generation principle of power batteries, a thermal model of power batteries is established:

[0056]

[0057] Among them, T s With T c Represent the battery surface temperature and battery core temperature respectively, T e Indicates ambient temperature; C s With C c Respectively represent the heat capacity coefficient of the battery internal material and the heat capacity coefficient of the battery surface; R c Represents the thermal resistance between the battery core and the surface; R u Represents the convection resistance between the battery surface and the cooling air; Q gen =I(U oc -U)=I(R o I+U1+U2) represents the heating power of the battery core, R o is the internal resistance of the power battery, U oc is the internal power supply of the battery, and U is the voltage across the battery;

[0058] Heat power Q gen and ambient temperature T e As input, the battery core temperature T c and surface temperature T s The power battery electrical model shown in formula (16) and the power battery thermal model shown in formula (17) are coupled and discretized to obtain the power battery electrical thermal model with unknown but bounded noise and Gaussian noise as shown in formula (1).

[0059] Optionally, the method further includes:

[0060] According to the power battery thermal model, the parameter matrices A and B are obtained, and their expressions are:

[0061]

[0062] Where Δt is the sampling time.

[0063] Optionally, the output matrix C is

[0064] The present application also provides the application of the above method in the production of power batteries.

[0065] The beneficial effects of the present invention are:

[0066] By obtaining the equivalent circuit model and thermal model of the power battery to establish an electric-thermal model of the power battery mixed with unknown but bounded noise and Gaussian noise, based on the Kalman filter concept, a polyhedral space update strategy is designed, and a state estimator is constructed. At the same time, the two evaluation indicators of minimum mean square error and minimum polyhedral space are considered to obtain the optimal weight of the dual filter, that is, the optimal weights of the two noises, noise that meets a certain probability distribution and unknown but bounded noise, are determined instead of taking the same weight, so that the estimated state interval is closer to the actual temperature of the power battery under the operating state, effectively improving the accuracy of the estimated interval containing the true value of the state. Compared with the existing state estimation method that combines random and unknown but bounded mixed noise dual filtering, the state estimation accuracy of the method of this application is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0068] Figure 1 It is a flow chart of a power battery temperature estimation method based on polyhedral filtering under double disturbance disclosed in one embodiment of the present invention.

[0069] Figure 2 It is a power battery equivalent circuit model diagram.

[0070] Figure 3 It is a comparison diagram of the upper and lower bounds of the core temperature of a power battery estimated by the present invention, the upper and lower bounds estimated by the existing method, and the true value of the core temperature when the power battery disclosed in one embodiment of the present invention is in a normal operating state.

[0071] Figure 4 It is a comparison diagram of the upper and lower bounds of the power battery surface temperature estimated by the present invention, the upper and lower bounds estimated by the existing method, and the surface true value when the power battery disclosed in one embodiment of the present invention is in normal operating state. DETAILED DESCRIPTION

[0072] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0073] Embodiment 1:

[0074] This embodiment provides a power battery temperature estimation method based on polyhedral filtering under double disturbance, see Figure 1 , the method comprising:

[0075] Step 1: Establish the electric thermal model of the power battery with unknown but bounded noise mixed with Gaussian noise:

[0076]

[0077] Among them, x k =[x 1,k ,x 2,k ] T =[T c,k ,T s,k ] T is the state variable, T c,k and T s,k They represent the battery core temperature and surface temperature at time k, respectively, and u k =[Q gen,k ,T e,k ] T is the input matrix at time k, Q gen Indicates the heat generation power of the battery core; T e,k represents the ambient temperature at time k, w k ∈<0,W> represents the unknown but bounded disturbance noise at time k, and W represents the generator matrix of the unknown but bounded disturbance noise; represents the disturbance noise that follows Gaussian distribution at time k, represents the covariance matrix of the disturbance noise that follows the Gaussian distribution; v k ∈<0,V> represents the unknown but bounded measurement noise at time k, and V represents the generator matrix of the unknown but bounded measurement noise; represents the measurement noise that obeys Gaussian distribution at time k, R represents the covariance matrix of the measurement noise that obeys Gaussian distribution; A represents the state space matrix, B represents the input matrix, and C represents the output matrix; y k is the output variable, i.e., the temperature measurement value of the power battery under the operating state;

[0078] Step 2: Based on the optimal estimate of the state at time k Determine the predicted value of the state at time k+1

[0079]

[0080]

[0081]

[0082] in, represents the predicted point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial prediction error, covariance matrix P k+1|kfor

[0083] P k+1|k =AP k A T +Q (5)

[0084] represents the prediction error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1|k > indicates that the generator matrix H k+1|k for:

[0085] H k+1|k =[AH k W] (6)

[0086] Step 3: Based on the predicted value of the state at time k+1 Determine the optimal estimate of the state at time k+1

[0087]

[0088]

[0089]

[0090] Where I is the identity matrix, y k+1 is the output of the system at time k+1, represents the point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial estimation error, covariance matrix P k+1 for:

[0091]

[0092] represents the estimated error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1 > indicates that the generator matrix H k+1 for:

[0093] H k+1 =[(IK k+1 C)H k+1|k K k+1 V] (11)

[0094] K k+1 is the optimal gain at time k+1. According to (10) and (11), the optimal criterion is:

[0095]

[0096] Among them, tr(Pk+1 ) indicates P k+1 traces, Indicates H k+1 The transpose of k+1 Represents the optimal weight coefficient at time k+1:

[0097]

[0098] Solve (12) to get the optimal gain K k+1 :

[0099]

[0100] Among them, M k+1 =[H k+1|k 0],

[0101] Step 4: Based on the optimal estimate of the state at time k+1 Determine the state estimation interval at time k+1 and complete the estimation of the power battery operating state:

[0102]

[0103] in, and They represent the upper and lower bounds of the Gaussian state estimation error confidence interval, and represent the upper and lower bounds of the unknown but bounded state estimation error interval, respectively.

[0104] Embodiment 2

[0105] This embodiment provides a power battery temperature estimation method based on polyhedral filtering under double disturbance, see Figure 1 , the method comprising:

[0106] Step 1: Establish a power battery equivalent circuit model based on the electrochemical mechanism of the power battery;

[0107] According to the electrochemical mechanism of power batteries, Figure 2 The power battery equivalent circuit shown in FIG. 1 is based on the power battery equivalent circuit, and the following equation is obtained:

[0108]

[0109] Among them, R1 and C p1 R1 and C are the electrochemical polarization internal resistance and capacitance of the power battery, respectively. p1 The voltage across the two terminals after they are connected in parallel is U1; R2 and C p2 R2 and C are the concentration polarization resistance and capacitance of the power battery respectively. p2After the two are connected in parallel, the voltage across the two ends is U2; I is the discharge current.

[0110] Step 2: Based on the thermal model of the power battery, an electric thermal model of the power battery with unknown but bounded noise mixed with Gaussian noise is established;

[0111] According to the heat generation principle of power batteries, a thermal model of power batteries is established:

[0112]

[0113] Among them, T s With T c Represent the battery surface temperature and battery core temperature respectively, T e Indicates ambient temperature; C s With C c Respectively represent the heat capacity coefficient of the battery internal material and the heat capacity coefficient of the battery surface; R c Represents the thermal resistance between the battery core and the surface; R u Represents the convection resistance between the battery surface and the cooling air; Q gen =I(U oc -U)=I(R o I+U1+U2) represents the heating power of the battery core, R o is the internal resistance of the power battery, U oc is the internal power supply of the battery, and U is the voltage across the battery.

[0114] Heat power Q gen and ambient temperature T e As input, the battery core temperature T c and surface temperature T s Assume that the power battery electrical model shown in formula (16) and the power battery thermal model shown in formula (17) are coupled and discretized to establish a power battery electrical thermal model with unknown but bounded noise and Gaussian noise mixed as follows:

[0115]

[0116] Among them, x k =[x 1,k ,x 2,k ] T =[T c,k ,T s,k ] T is the state variable, T c,k and T s,k They represent the battery core temperature and surface temperature at time k, respectively, and u k =[Q gen,k ,T e,k ] T is the input matrix at time k, Qgen Indicates the heat generation power of the battery core; T e,k represents the ambient temperature at time k, w k ∈<0,W> represents the unknown but bounded disturbance noise at time k, represents the disturbance noise that follows Gaussian distribution at time k; v k ∈<0,V> represents the unknown but bounded measurement noise at time k, represents the measurement noise that follows Gaussian distribution at time k; A represents the state space matrix, B represents the input matrix, C represents the output matrix, y k is the output variable, is the parameter matrix of the system, and k is the discrete time.

[0117] Step 3: Based on the optimal estimate of the state at time k Determine the predicted value of the state at time k+1

[0118]

[0119]

[0120]

[0121] in, represents the predicted point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial prediction error, covariance matrix P k+1|k for

[0122] P k+1|k =AP k A T +Q (5)

[0123] represents the prediction error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1|k > indicates that the generator matrix H k+1|k for:

[0124] H k+1|k =[AH k W] (6)

[0125] Step 4: Based on the predicted value of the state at time k+1 Determine the optimal estimate of the state at time k+1

[0126]

[0127]

[0128]

[0129] Where I is the identity matrix, y k+1 is the output of the system at time k+1, represents the point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial estimation error, covariance matrix P k+1 for:

[0130]

[0131] represents the estimated error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1 > indicates that the generator matrix H k+1 for:

[0132] H k+1 =[(IK k+1 C)H k+1|k K k+1 V] (11)

[0133] K k+1 is the optimal gain at time k+1. According to (10) and (11), the optimal criterion is:

[0134]

[0135] Among them, tr(P k+1 ) indicates P k+1 traces, Indicates H k+1 The transpose of k+1 Represents the optimal weight coefficient at time k+1:

[0136]

[0137] Solve (12) to get the optimal gain K k+1 :

[0138]

[0139] Among them, M k+1 =[H k+1|k 0],

[0140] The optimal gain K k+1 Substitute into formula (8) to obtain the point estimate of the state at time k+1 under the optimal gain;

[0141] The optimal gain K k+1 Substituting into formula (10) and (11) respectively, we can obtain the Gaussian distribution at time k+1 under the optimal gain: The estimation error of the part, and the estimation error of the unknown but bounded part at time k+1 under the optimal gain;

[0142] Then, the optimal estimate is calculated according to formula (7):

[0143] Step 5: Consider the influence of unknown but bounded noise and set the error tolerance interval. Determine the state estimation interval at time k+1 and complete the estimation of the power battery operating state:

[0144]

[0145] in, and They represent the upper and lower bounds of the Gaussian state estimation error confidence interval, and represent the upper and lower bounds of the unknown but bounded state estimation error interval, respectively.

[0146] In this example, after executing steps 1 to 5 within a predetermined time range, a state estimation interval is obtained to achieve estimation of the power battery operating state.

[0147] Figure 3 and Figure 4 The estimated interval change diagram of the core temperature state variable and the estimated interval change diagram of the surface temperature state variable of the power battery are respectively obtained by using the method provided by this application and the existing method (the existing method can refer to An extended zonotopic and Gaussian Kalman filter (EZGKF) merging set-membership and stochastic paradigms: Toward non-linear filtering and fault detection). Among them, Figure 3 For comparison of the estimation results of the power battery core temperature; Figure 4 Comparison of the estimated results of the power battery surface temperature.

[0148] from Figure 3 and Figure 4It can be seen that the existing estimation method and the estimation method proposed in the present application can both achieve state estimation of the core temperature and surface temperature of the power battery, and the true value of the state is within the estimation interval. However, the method proposed in the present application determines the optimal weights for the two types of noise, namely, noise that satisfies a certain probability distribution and unknown but bounded noise, instead of taking the same weights. Therefore, the estimated temperature value interval is smaller and more accurate.

[0149] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0150] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A power battery temperature estimation method based on polyhedral filtering under double disturbance, characterized in that: The method comprises: Step 1: Establish a power battery electrothermal model with a mixture of unknown but bounded noise and Gaussian noise; Step 2: Based on the optimal estimate of the temperature at time k Determine the predicted value of the temperature at time k+1 Consider the corresponding prediction errors for unknown but bounded noise and Gaussian noise respectively; Step 3: Based on the predicted value of temperature at time k+1 Determine the best estimate of the temperature at time k+1 Consider the corresponding estimation errors for unknown but bounded noise and Gaussian noise respectively and determine the optimal weight coefficient at time k+1 to achieve a trade-off between the two noises; Step 4: Consider the influence of unknown but bounded noise and calculate the optimal estimate of the temperature at time k+1 Obtain the temperature estimation interval at time k+1 to complete the temperature estimation of the power battery under the operating state; The Step 2 includes: If k=0, the optimal estimated value at the initial moment is set according to the core temperature and surface temperature of the power battery at the initial operation moment, and then the predicted value of the temperature at the next moment is determined according to the optimal estimated value of the temperature at the initial moment; If k≠0, the optimal estimated value at time k is set according to the core temperature and surface temperature of the power battery at time k, and then the optimal estimated value of the temperature at time k is set according to the optimal estimated value of the temperature at time k. Determine the predicted value of the temperature at time k+1 in, represents the predicted point estimate of the temperature at time k+1, Indicates that k+1 time follows Gaussian distribution Partial prediction error, covariance matrix P k+1|k for: P k+1|k =AP k From T +Q (5) Among them, P k is the covariance matrix of the estimation error that obeys Gaussian distribution at time k; Q is the covariance matrix of the disturbance noise that obeys Gaussian distribution; represents the prediction error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1k > indicates that the generator matrix H k+1|k for: H k+1|k =[AH k W] (6) Among them, H k is the generator matrix of the unknown but bounded part of the estimation error at time k; W is the generator matrix of the unknown but bounded disturbance noise.

2. The method according to claim 1, characterized in that The power battery electrothermal model of unknown but bounded noise mixed with Gaussian noise established in Step 1 is: Among them, x k =[x 1,k ,x 2,k ] T =[T c,k ,T s,k ] T is the temperature variable, T c,k and T s,k They represent the battery core temperature and surface temperature at time k, respectively, and u k =[Q gen,k ,T e,k ] T is the input matrix at time k, Q gen Indicates the heat generation power of the battery core; T e,k represents the ambient temperature at time k, w k ∈<0,W> represents the unknown but bounded disturbance noise at time k, represents the disturbance noise that follows Gaussian distribution at time k; v k ∈<0,V> represents the unknown but bounded measurement noise at time k, represents the measurement noise that follows Gaussian distribution at time k; A represents the state space matrix, B represents the input matrix, and C represents the output matrix; y k is the output variable, i.e. the temperature measurement value of the power battery under the operating state.

3. The method according to claim 2, characterized in that The Step 3 includes: The optimal estimate of the state at time k+1 is calculated according to the following formula (7): Where I is the identity matrix, y k+1 is the output of the system at time k+1, represents the point estimate of the state at time k+1, Indicates that k+1 time follows Gaussian distribution Partial estimation error, covariance matrix P k+1 for: C is the output matrix; R is the covariance matrix of the measurement noise that obeys the Gaussian distribution; represents the estimated error of the unknown but bounded part at time k+1, which is represented by the fully symmetric polytope <0,H k+1 > indicates that the generator matrix H k+1 for: H k+1 =[(I-K k+1 C)H k+1|k K k+1 V] (11) V is the generator matrix of the unknown but bounded measurement noise; K k+1 is the optimal gain at time k+1. According to (10) and (11), the optimal criterion is: Among them, tr(P k+1 ) indicates P k+1 traces, Indicates H k+1 The transpose of k+1 Represents the optimal weight coefficient at time k+1: Solve the optimal criterion shown in equation (12) to obtain the optimal gain K k+1 : Among them, M k+1 =[H k+1|k 0], The optimal gain K k+1 Substitute into formula (8) to obtain the point estimate of the state at time k+1 under the optimal gain; The optimal gain K k+1 Substituting into formula (10) and (11) respectively, we can obtain that the optimal gain at time k+1 follows the Gaussian distribution The estimation error of the part, and the estimation error of the unknown but bounded part at time k+1 under the optimal gain; Then, the optimal estimate is calculated according to formula (7):

4. The method according to claim 3, characterized in that: The Step 4 includes: Considering the influence of unknown but bounded noise, setting the tolerance interval, according to the optimal estimate of the temperature at time k+1 Determine the temperature estimation interval at time k+1 and complete the temperature estimation of the power battery under the operating state in, and They represent the upper and lower bounds of the Gaussian state estimation error confidence interval, and represent the upper and lower bounds of the unknown but bounded state estimation error interval, respectively.

5. The method according to claim 4, characterized in that The Step 1 includes: According to the electrochemical mechanism of the power battery, an equivalent circuit of the power battery is established. According to the equivalent circuit of the power battery, the following equation is obtained: Among them, R1 and C p1 R1 and C are the electrochemical polarization internal resistance and capacitance of the power battery, respectively. p1 The voltage across the two terminals after they are connected in parallel is U1; R2 and C p2 R2 and C are the concentration polarization resistance and capacitance of the power battery respectively. p2 The voltage across the two terminals after they are connected in parallel is U2; I is the discharge current; According to the thermal model of the power battery, an electrothermal model of the power battery with unknown but bounded noise mixed with Gaussian noise is established; According to the heat generation principle of power batteries, a thermal model of power batteries is established: Among them, T s With T c Represent the battery surface temperature and battery core temperature respectively, T e Indicates ambient temperature; C s With C c Respectively represent the heat capacity coefficient of the battery internal material and the heat capacity coefficient of the battery surface; R c Represents the thermal resistance between the battery core and the surface; R u Represents the convection resistance between the battery surface and the cooling air; Q gen =I(U oc -U)=I(R o I+U1+U2) represents the heating power of the battery core, R o is the internal resistance of the power battery, U oc is the internal power supply of the battery, and U is the voltage across the battery; Heat power Q gen and ambient temperature T e As input, the battery core temperature T c and surface temperature T s The power battery electrical model shown in formula (16) and the power battery thermal model shown in formula (17) are coupled and discretized to obtain the power battery electrical thermal model with unknown but bounded noise and Gaussian noise as shown in formula (1).

6. The method according to claim 5, characterized in that The method further comprises: According to the power battery thermal model, the parameter matrices A and B are obtained, and their expressions are: Where Δt is the sampling time.

7. The method according to claim 6, characterized in that The output matrix C is 8. Application of the method according to any one of claims 1 to 7 in power battery production.

Images