Patent Application
20130116954
The present invention relates to a battery state estimating device and a battery state estimating method for estimating a state inside a secondary battery.
A control device described below is known as a control device of a secondary battery. The control device defines a predetermined battery model, estimates a terminal voltage of the secondary battery which is based on the battery model, as an estimated voltage value on the basis of measured values of a current and a terminal voltage of the secondary battery, sequentially identifies parameters of the battery model in such a way that the difference between the measured voltage value and the estimated voltage value converges to zero, and thereby estimates the state inside the battery (see, for example, Patent Literature 1).
In the related art described above, there is a case where the sequentially-identified parameters of the battery model include identification errors. The identification error of a certain parameter φn among the sequentially-identified parameters of the battery model greatly affects the estimation result of the state inside the battery. Accordingly, when the identification error of the certain parameter φn is large, the estimation of the state inside the battery cannot be performed properly in some cases.
An object of the present invention is to provide a battery state estimating device and a battery state estimating method capable of properly estimating a state inside a secondary battery.
One embodiment of the present invention is a battery state estimating device as follows. The battery state estimating device detects a current and a terminal voltage of a secondary battery, estimates a terminal voltage of the secondary battery which is based on a predetermined battery model by using the detected measured values of the current and the terminal voltage, and sequentially identifies parameters of the battery model in such a way that the difference between a value based on the measured value of the terminal voltage and the estimate value of the terminal voltage converges to zero. In the battery state estimating device, when a certain parameter among the identified parameters is equal to or larger than a predetermined first threshold, upper limit processing of setting a value of the certain parameter to the first threshold is performed and the problem described above is thereby solved.
In one embodiment of the present invention, when the certain parameter among the identified parameters is equal to or larger than the predetermined first threshold, the value of the certain parameter is set to the first threshold. This reduces an estimation error of the state inside battery which is caused by an identification error of the certain parameter. Accordingly, the state inside the secondary battery can be properly estimated.
One or more embodiments of the present invention are described below based on the drawings.
A secondary battery 10 is formed by connecting multiple unit cells in series. Lithium-based secondary batteries such as lithium-ion secondary batteries can be given as an example of the unit cells forming the secondary battery 10. A motor can be given as an example of a load 20.
A current sensor 40 is a sensor configured to detect charge and discharge currents flowing through the secondary battery 10. A signal detected by the current sensor 40 is transmitted to an electronic control unit 30. Moreover, a voltage sensor 50 is a sensor configured to detect a terminal voltage of the secondary battery 10. A signal detected by the voltage sensor 50 is transmitted to the electronic control unit 30. A temperature sensor 60 for detecting the temperature of the secondary battery 10 is provided near the secondary battery 10. The temperature sensor 60 is a sensor using a thermocouple. A signal detected by the temperature sensor 60 is also similarly transmitted to the electronic control unit 30.
The electronic control unit 30 is a control unit for controlling the secondary battery 10 and includes a microcomputer, an electronic circuit, and the like, the microcomputer including a CPU configured to execute calculation processing specified in a computer program and ROM and RAM configured to store the computer program and calculation results.
As shown in
The current detecting part 301 acquires the signal from the ammeter 40 at predetermined intervals and detects the charge or discharge current flowing through the secondary battery 10, on the basis of the signal from the ammeter 40, to acquire a measured current value I(k). The current detecting part 301 transmits the acquired measured current value I(k) to the state variable filter calculation part 304.
The voltage detecting part 302 acquires the signal from the voltmeter 50 at predetermined intervals and detects the terminal voltage of the secondary battery 10 on the basis of the signal from the voltmeter 50 to acquire a measured voltage value V(k). The voltage detecting part 302 transmits the acquired measured voltage value V(k) to the state variable filter calculation part 304 and the adaptive identification calculation part 305.
The temperature detecting part 303 acquires the signal from the temperature sensor 60 at predetermined intervals and detects the temperature of the secondary battery 10 on the basis of the signal from the temperature sensor 60 to acquire a battery temperature T(k). The temperature detecting part 303 transmits the acquired battery temperature T(k) to the adaptive identification calculation part 305 and the open circuit voltage estimating part 306.
The state variable filter calculation part 304 defines a battery model of the secondary battery 10, performs state variable filter calculation by using the measured current value I(k) detected by the current detecting part 301 and the measured voltage value V(k) detected by the voltage detecting part 302, and thereby obtains a transformed state amount ω(k).
A method of calculating the transformed state amount ω(k) which is performed by the state variable filter calculation part 304 is described below.
First, the “battery model” used in the embodiment is described.
In this formula, a model input is a current I [A] and a model output is a terminal voltage V [V]. Moreover, R1 [Ω] is a charge transfer resistance, R2 [Ω] is a pure resistance, C1 [F] is an electrical double-layer capacitance, V0 [V] is an open circuit voltage. The current I [A] being a positive value indicates charge and the current I [A] being a negative value indicates discharge. Moreover, s in Formula (1) shown above is a differential operator. Although the battery model of the embodiment is a reduction model (linear function) in which a positive electrode and a negative electrode are not particularly separated from each other, the battery model can relatively accurately show the actual charge and discharge characteristics of the battery. As described above, in the embodiment, description is given of an example of a configuration in which the degree of the battery model is one. The degree of the battery model may be two or more.
When R1, R2, and C1 are expressed as in Formula (2) shown below, Formula (1) shown above is expressed as in Formula (3) shown below.
Then, in the embodiment, the state variable filter calculation part 304 performs the state variable filter calculation on the basis of the battery model expressed in Formula (3) shown above to obtain the transformed state amount ω(k).
First, assuming that the open circuit voltage V0(t) is a value obtained by integrating a certain value, which is obtained by multiplying the current I(t) by a variable parameter h, from a certain initial state, the open circuit voltage V0(t) can be expressed by Formula (4) shown below.
Subsequently, Formula (4) shown above is substituted into Formula (3) shown above and Formula (5) shown below is thereby obtained. Formula (5) is then simplified to Formula (6) shown below.
Formula (6) shown above corresponds to a formula where degrees of A(s) and B(s) are respectively one and two in Formula (7) shown below.
In this formula, A(s) and B(s) are polynomial functions of s.
In Formula (6) shown above, (T1·s2+s) being the denominator on the right side corresponds to s·A(s) of Formula (7) shown above and Formula (8) shown below is obtained by dividing the denominator and the numerator on the right side of Formula (6) shown above by T1 being the coefficient of a term with the highest degree in (T1·s2+s). Note that Formula (8) shown below is a formula where the coefficient of s2 being the term with the highest degree in the denominator on the right side of Formula (6) shown above is one.
When T1 being an unknown parameter is the coefficient of s2 being the term with the highest degree in the denominator on the right side as in Formula (6) shown above, there may be a case where T1 being the unknown parameter is calculated to be a value close to zero in an identification step of the unknown parameter to be described later. In such a case, the degree of the denominator on the right side of Formula (6) shown above may change. When the degree of the denominator on the right side of Formula (6) changes, the following problems may occur. A delay in identification of unknown parameter to be described later occurs or the unknown parameter does not converge to a true value after the identification calculation. In the embodiment, the denominator and the numerator on the right side of Formula (6) shown above are divided by T1 and Formula (8) shown above is thereby obtained to effectively prevent the problems described above.
Then, a known fixed number ki (i=1, 2, . . . , n) is introduced into Formula (8) shown above and Formulae (9) and (10) shown below can be thereby obtained.
In Formula (10) shown above, y(t) is a value obtained by subtracting a direct feedthrough term from V(t). Accordingly, y(t) is expressed by Formula (11) shown below.
In Formula (10) shown above, Ii and b0i are parameters including unknown parameters (T1, T2, K, h) and fVi and fIi are transformed state amounts obtained by subjecting I(k) and V(k), which are values measurable by the ammeter 40 and the voltmeter 50, to filter processing using the state variable filter. Since Formula (10) shown above is a formula of a sum of products of Ii, b0i, fVi, and fIi, Formula (10) coincides with Formula (12) shown below being a normal form of an adaptive digital filter.
[Formula 12]
y(t)=φTω (12)
Note that, in Formula (12) shown above, φT=[Ii, b0i] and ω=[fVi, fIi] are satisfied.
The transformed state amount ω(k) is thus calculated by the state variable filter calculation part 304. The state variable filter calculation part 304 then transmits the obtained transformed state amount ω(k) to the adaptive identification calculation part 305 and the open circuit voltage estimating part 306.
The adaptive identification calculation part 305 performs adaptive digital filter calculation on the basis of the transformed state amount ω(k) calculated by the state variable filter calculation part 304 and thereby performs the identification calculation of identifying parameters φ̂(k) of the battery model of the secondary battery 10.
Note that “̂” attached to an upper right portion in φ̂(k) indicates that the value to which “̂” attached is an estimate value. Moreover, although “̂” indicating the estimate values are provided directly above “φ” of φ(k), “V” of V0(k), and “S” of SOC(k) in
[Formula 13]
φ̂(k)={circumflex over (φ)}(k)
V
0̂(k)={circumflex over (V)}0(k)
SOĈ(k)=ŜOC(k) (13)
Specifically, in a configuration diagram of an adaptive identification system shown in
Then, the adaptive identification calculation part 305 performs the identification calculation of identifying the parameter φ̂(k) of the battery model on the basis of an algorithm shown in Formula (14) shown below, by using an adaptive adjustment scheme. In the identification calculation, the adaptive identification calculation part 305 performs the identification calculation in such a way that the difference between the estimated voltage value V̂(k) and the measured voltage value V(k) being an actual measured value which is detected by the voltmeter 50 and acquired by the voltage detecting part 302 converges to zero. In the embodiment, “both-limit trace gain method” in which logical disadvantages of a simple “adaptive digital filter using a least squares method” are improved may be used in the calculation. The following point can be given as an example of the logical disadvantages. Once the estimate value converges, accurate estimation cannot be performed again when the parameters change after the convergence. Note that
Formula (14) shown above is a sequential formula for adaptively obtaining the parameter φ̂(k). Furthermore, γ(k) and Γ(k−1) are both adaptive gains. Out of these, γ(k) is a scalar gain (error gain) and Γ(k−1) is a matrix gain (signal gain). When the transformed state amount ω(k) at a time point k is obtained from Formula (14) shown above, e(k) can be obtained, e(k) being the difference between the estimated voltage value V̂(k) being the estimate value of the terminal voltage of the secondary battery 10 which is estimated from the battery model and the measured voltage value V(k) detected by the voltmeter 50 and acquired by the voltage detecting part 302. Then, converging e(k) to zero allows the parameter φ̂(k) to be sequentially calculated by using the parameter φ̂(k−1) and the matrix gain Γ(k−1) which are obtained in the previous processing. In this case, it is possible to subject the value acquired by the voltage detecting part 302 to filter processing for removal of measurement noise and to use the resultant value as the measured voltage value V(k). In such a case, the estimated voltage value taking filter characteristics into consideration is calculated.
The parameter φ̂(k) of the secondary battery 10 which is calculated as described above is transmitted from the adaptive identification calculation part 305 to the open circuit voltage estimating part 306 as shown in
Then, as shown in
Subsequently, the adaptive identification calculation part 305 calculates the estimated voltage value V̂(k) being the estimate value of the terminal voltage which is based on the battery model, on the basis of the transformed state amount ω(k) obtained by the state variable filter calculation part 304 and the parameter φ̂(k−1) of the battery model. Then, in the adaptive identification system of the embodiment, the adaptive identification calculation part 305 performs the sequential identification of the parameter φ̂(k) of the battery model in accordance with Formula (14) shown above, by using the transformed state amount ω(k) obtained by the state variable filter calculation part 304, the measured voltage value V(k) detected by the voltage detecting part 302, and the estimated voltage value V̂(k). Note that the parameter φ̂(k) of the battery model is generally a parameter vector including multiple parameters. In the example shown in
When T1 being the coefficient of the term with the highest degree in (T1·s2+s) being the denominator on the right side of Formula (6) shown above is expressed by using the parameter φ̂(k) of the battery model which is identified by the adaptive identification calculation part 305, Formula (15) shown below is obtained.
In Formula (15) shown above, φn is a highest-degree term parameter corresponding to T1 being the coefficient of the term with the highest degree in (T1·s2+s) being the denominator on the right side of Formula (6) shown above, among the parameters included in the parameter φ̂(k) of the battery model. In other words, the highest-degree term parameter φn is a parameter corresponding to T1 in the parameter φ̂(k) of the battery model which is identified by the adaptive identification calculation part 305. Moreover, α is a value determined depending on a cutoff frequency λ of the state variable filter.
Then, the adaptive identification calculation part 305 determines whether the highest-degree term parameter φn corresponding to T1 being the coefficient of the term with the highest degree is equal to or larger than an initialize determination threshold δ2. In other words, the adaptive identification calculation part 305 determines whether the condition of “φn≧δ2” is satisfied. Here, the initialize determination threshold δ2 is a threshold for determining whether the identification calculation of the parameter φ̂(k) of the battery model which is performed by the adaptive identification calculation part 305 is to be initialized. The initialize determination threshold δ2 is calculated from the battery temperature T(k) detected by the temperature detecting part 303, by using a table which is stored in advance in the ROM included in the electronic control unit 30 and which shows the relationship between the battery temperature and the initialize determination threshold δ2. Moreover, the initialize determination threshold δ2 is set to a value larger than an upper-limit determination threshold δ1 to be described later. In other words, “initialize determination threshold δ2>upper-limit determination threshold δ1” is satisfied.
When the adaptive identification calculation part 305 determines whether the condition of “φn≧δ2” is satisfied and consequently determines that the condition of “φn≧δ2” is satisfied, it is judged that the value of the parameter φ̂(k) identified in the sequential identification calculation using Formula (14) shown above greatly changes and the identified parameter φ̂(k) thereby greatly deviates from the true value. Hence, initialization processing of the identification calculation is performed. Specifically, when the condition of “φn≧δ2” is satisfied, the adaptive identification calculation part 305 sets the parameter φ̂(k−1) and the matrix gain Γ(k−1) which are used in the next calculation respectively to a parameter initial value φi and a matrix gain initial value Γi, upon performing the sequential identification calculation by using Formula (14) shown above. Such initialization processing allows the parameter φ̂(k) to be set close to the true value again in a relatively short time even when the identified parameter φ̂(k) greatly deviates from the true value.
The open circuit voltage estimating part 306 estimates the open circuit voltage of the secondary battery 10 on the basis of the transformed state amount ω(k) obtained by the state variable filter calculation part 304 and the parameter φ̂(k) obtained by the adaptive identification calculation part 305, and thereby calculates the estimate open circuit voltage value V0̂(k).
Specifically, the open circuit voltage estimating part 306 calculates Formula (16) shown below on the basis of Formula (3) shown above and calculates the estimate open circuit voltage value V0̂(k) from Formula (16) shown below, by using the parameter φ̂(k) calculated from Formula (14) shown above, the transformed state amount ω(k) calculated based on Formula (10) shown above, and the cutoff frequency λ of the state variable filter.
In Formula (16) shown above, φn is the highest-degree term parameter corresponding to T1 being the coefficient of the term with the highest degree. Moreover, α is a value determined depending on the cutoff frequency λ of the state variable filter.
Furthermore, the open circuit voltage estimating part 306 determines whether the highest-degree term parameter φn corresponding to T1 being the coefficient of the term with the highest degree is equal to or larger than the upper-limit determination threshold δ1. In other words, the open circuit voltage estimating part 306 determines whether the condition of “φn≧δ1” is satisfied. Here, the upper-limit determination threshold δ1 is an upper limit value of the highest-degree term parameter φn which is used when the estimate open circuit voltage value V0̂(k) is calculated by the open circuit voltage estimating part 306. The upper-limit determination threshold δ1 is calculated from the battery temperature T(k) detected by the temperature detecting part 303, by using a table which is stored in advance in the ROM included in the electronic control unit 30 and which shows the relationship between the battery temperature and the upper-limit determination threshold δ1.
When the open circuit voltage estimating part 306 determines whether the condition of “φn≧δ1” is satisfied and consequently determines that the condition of “φn≧δ1” is satisfied, the open circuit voltage estimating part 306 performs upper limit processing of setting the value of the highest-degree term parameter φn to a value equal to δ1 (hereafter, this value will be referred to as upper limit value φn
In the calculation, when the highest-degree term parameter φn becomes a value close to the value α determined depending on the cutoff frequency λ of the state variable filter in Formula (16) shown above, (α−φn) being a component of the denominator of Formula (16) shown above becomes close to zero. Consequently, the identification error of the highest-degree term parameter φn largely affects the estimation error of the estimate open circuit voltage value V0̂(k). In the embodiment, such a problem is effectively solved by setting the upper limit value φn
The open circuit voltage estimating part 306 can calculate the unknown parameters (T1, T2, K, and h) in addition to the estimate open circuit voltage value V0̂(k) from Formula (16) shown above.
The SOC estimating part 307 calculates an estimate state-of-charge value SOĈ(k) from the estimate open circuit voltage value V0̂(k) calculated by the open circuit voltage estimating part 306, on the basis of a predetermined open circuit voltage-state of charge characteristic of the secondary battery 10.
Next, estimation processing of the parameter φ̂(k) and the estimate state-of-charge value SOĈ(k) in the embodiment is described by using the flowchart shown in
In Step S1, the current detecting part 301, the voltage detecting part 302, and the temperature detecting part 303 acquire the measured current value I(k), the measured voltage value V(k), and the battery temperature T(k), respectively. Then, the measured current value I(k) is transmitted to the state variable filter calculation part 304, the measured voltage value V(k) is transmitted to the state variable filter calculation part 304 and the adaptive identification calculation part 305, and the battery temperature T(k) is transmitted to the adaptive identification calculation part 305 and the open circuit voltage estimating part 306.
In Step S2, the state variable filter calculation part 304 performs the state filter calculation using the state variable filter on the basis of Formulae (9) and (10) shown above, for the measured current value I(k) and the measured voltage value V(k) which are acquired in Step S1. The transformed state amount ω(k) is thereby calculated.
In Step S3, the calculation of the upper-limit determination threshold δ1 and the initialize determination threshold δ2 is performed based on the battery temperature T(k) acquired in Step S1. Specifically, the upper-limit determination threshold δ1 is calculated by the open circuit voltage estimating part 306 from the battery temperature T(k) acquired in Step S1, by using the table which is stored in advance in the ROM included in the electronic control unit 30 and which shows the relationship between the battery temperature and the upper-limit determination threshold δ1. Moreover, the initialize determination threshold δ2 is calculated by the adaptive identification calculation part 305 from the battery temperature T(k) acquired in Step S1, by using the table which is stored in advance in the ROM included in the electronic control unit 30 and which shows the relationship between the battery temperature and the initialize determination threshold δ2.
In a secondary battery such as a lithium-ion secondary battery, a diffusion resistance (corresponding to the charge transfer resistance R1) tends to become larger as the battery temperature T(k) becomes lower. Accordingly, T1 being the coefficient of the term with the highest degree tends to be calculated as a larger value as the battery temperature T(k) becomes lower (see Formula (2) shown above). From Formulae (15) and (16) shown above, the fact that T1 being the coefficient of the term with the highest degree tends to be calculated as a larger value means that (α−φn) being a component of the denominator in Formulae (15) and (16) shown above tends to be calculated as a value close to zero. To counter this, in the embodiment, the lower the battery temperature T(k) is, the larger the values of the upper-limit determination threshold δ1 and the initialize determination threshold δ2 are set. In other words, the upper-limit determination threshold δ1 and the initialize determination threshold δ2 are set to values close to the value α determined depending on the cutoff frequency λ of the state variable filter. This enables the later-described initialization processing in Step S5 and the later-described upper limit processing in Step S8 to be performed properly depending on the battery temperature T(k). Excellent identification accuracy of the parameter φ̂(k) and excellent estimation accuracy of the estimate open circuit voltage value V0̂(k) and the estimate state-of-charge value SOĈ(k) are thereby achieved.
In Step S4, the adaptive identification calculation part 305 determines whether the highest-degree term parameter φn(k−1) among the parameters included in the parameter φ̂(k−1) calculated in Step S6 of the previous processing is equal to or larger than the initialize determination threshold δ2 calculated in Step S3. In other words, in Step S4, determination of whether “φn(k−1)≧δ2” is satisfied is performed. When “φn(k−1)≧δ2” is satisfied, the processing proceeds to Step S5. Meanwhile, when “φn(k−1)≧δ2” is not satisfied, the processing proceeds to Step S6.
When the adaptive identification calculation part 305 determines that “φn(k−1)≧δ2” is satisfied in Step S4, the processing proceeds to Step S5. In Step S5, the adaptive identification calculation part 305 performs the initialization processing in which the value of the parameter φ̂(k−1) and the value of the matrix gain Γ(k−1) which are used in the identification calculation performed based on Formula (14) shown above are set respectively to the predetermined parameter initial value φi and the predetermined matrix gain initial value Γi.
In Step S6, the adaptive identification calculation part 305 performs the identification calculation processing for identifying the parameter φ̂(k) of the battery model. Specifically, the adaptive identification calculation part 305 first estimates the estimated voltage value V̂(k) of the secondary battery 10 which is estimated from the battery model, from the transformed state amount ω(k) calculated in Step S2. Then, the adaptive identification calculation part 305 performs the identification calculation of identifying the parameter φ̂(k) of the battery model on the basis of the algorithm shown in Formula (14) shown above, by using the adaptive adjustment scheme. In this case, the adaptive identification calculation part 305 performs the identification calculation described above in such a way that the difference between the estimated voltage value V̂(k) and the measured voltage value V(k) being the actual measured value detected by the voltmeter 50 and acquired by the voltage detecting part 302 converges to zero. Note that, when the parameter φ̂(k−1) and the matrix gain Γ(k−1) are set respectively to the parameter initial value φi and the matrix gain initial value Γi in Step S5, the calculation of the parameter φ̂(k) of the battery model is performed by using these initial values. Then, the parameter φ̂(k) of the battery model which is calculated by the adaptive identification calculation part 305 is transmitted to the open circuit voltage estimating part 306.
In Step S7, the open circuit voltage estimating part 306 determines whether the highest-degree term parameter φn(k) among the parameters included in the parameter φ̂(k) of the battery model which is calculated in Step S6 of the latest processing is equal to or larger than the upper-limit determination threshold δ1 calculated in Step S3. In other words, in Step S7, determination of whether “φn(k)≧δ1” is satisfied is performed. When “φn(k)≧δ1” is satisfied, the processing proceeds to Step S8. Meanwhile, when “φn(k)≧δ1” is not satisfied, the processing proceeds to Step S9.
When the open circuit voltage estimating part 306 determines that “φn(k)≧δ1” is satisfied is Step S7, the processing proceeds to Step S8. In Step S8, the open circuit voltage estimating part 306 performs the upper limit processing of setting the value of the highest-degree term parameter φn to the value equal to δ1, i.e. the upper limit value φn
Next, in Step S9, the open circuit voltage estimating part 306 calculates the estimate open circuit voltage value V0̂(k) on the basis of Formula (16) shown above, by using the parameter φ̂(k) of the battery model calculated in Step S6, the transformed state amount ω(k) calculated in Step S2, and the cutoff frequency λ of the state variable filter. In the calculation of the estimate open circuit voltage value V0̂(k), when the upper limit processing of setting the value of the highest-degree term parameter φn to the upper limit value φn
In Step S10, the SOC estimating part 307 performs the calculation of the estimate state-of-charge value SOĈ(k) on the basis of the predetermined open circuit voltage-state of charge characteristic of the secondary battery 10, by using the estimate open circuit voltage value V0̂(k) calculated in Step S9. Thereafter, the processing returns to Step S1. The processing of Step S1 to Step S10 described above is then repeated and the estimation processing of the parameter φ̂(k) of the battery model and the estimation processing of the estimate state-of-charge value SOĈ(k) which is based on the parameter φ̂(k) are thereby repeatedly performed.
In the embodiment, when the highest-degree term parameter φn corresponding to T1 being the coefficient of the term with the highest degree in (T1·s2+s) being the denominator on the right side of Formula (6) shown above in the parameter φ̂(k) of the battery model is equal to or larger than the upper-limit determination threshold δ1, the upper limit processing of setting the value of the highest-degree term parameter φn to the value equal to δ1 is performed. Then, the estimation of the estimate open circuit voltage value V0̂(k) is performed according to Formula (16) shown above, by using the value subjected to the upper limit processing as the highest-degree term parameter φn. Accordingly, in the embodiment, it is possible to effectively prevent the case where (α−n) being the component of the denominator in Formula (16) shown above becomes close to zero and the estimation error of the estimate open circuit voltage value V0̂(k) thereby becomes large. Accordingly, the estimation accuracy of the estimate open circuit voltage value V0̂(k) and the estimate state-of-charge value SOĈ(k) can be improved.
Moreover, in the embodiment, when the upper limit processing is performed for the highest-degree term parameter φn, the value subjected to the upper limit processing is used in the calculation of the estimate open circuit voltage value V0̂(k) which is performed by the open circuit voltage estimating part 306 as described above. Meanwhile, when the sequential identification calculation is performed according to Formula (14) shown above by the adaptive identification calculation part 305, instead of the value subjected to the upper limit processing, the value identified in the identification calculation processing is used as it is as the highest-degree term parameter φn(k−1) used in the next calculation. In other words, the configuration of the embodiment is such that the value subjected to the upper limit processing is not used when the sequential identification calculation is performed according to Formula (14) shown above by the adaptive identification calculation part 305. Using the value identified in the identification calculation processing as it is instead of the value subjected to upper limit processing in the sequential identification calculation enables the adaptive identification calculation part 305 to perform sequential calculation for the identification of the parameter φ̂(k) of the battery model in an ideal linear state. Hence, the identification accuracy of the parameter φ̂(k) of the battery model can be maintained at an excellent level. Particularly, when the value subjected to the upper limit processing is used in the sequential identification calculation performed by the adaptive identification calculation part 305, the non-linear state may occur in some cases. In such a case, there may be a problem of the sequential identification calculation not being performed at an excellent level, such as a problem of the convergence to the true value taking time. Such a problem can be effectively prevented by not using the value subjected to the upper limit processing when the sequential identification calculation is performed according to Formula (14) shown above by the adaptive identification calculation part 305.
Although the embodiment shows an example of the configuration in which, when the sequential identification calculation is performed by the adaptive identification calculation part 305, instead of the value subjected to the upper limit processing, the value identified in the identification calculation processing is used as it is as the highest-degree term parameter φn(k−1), the invention is not particularly limited to this configuration. The configuration may be such that the value subjected to the upper limit processing is used. However, as described above, when there is a possibility of non-linear state occurring due to the use of the value subjected to upper limit processing, the identification accuracy of the parameter φ̂(k) of the battery model can be improved by using the value identified in the identification calculation processing as it is, instead of the value subjected to the upper limit processing.
Moreover, in the embodiment, when the highest-degree term parameter φn is equal to or larger than the initialize determination threshold δ2 larger than the upper-limit determination threshold δ1, it is judged that the value of the parameter φ̂(k) identified in the sequential identification calculation of the parameter φ̂(k) of the battery model which is performed by the adaptive identification calculation part 305 greatly changes and the identified parameter φ̂(k) thereby greatly deviates from the true value. Hence, the adaptive identification calculation part 305 initializes the sequential identification calculation of the parameter φ̂(k) of the battery model. In the embodiment, this initialization allows the parameter φ̂(k) to be set close to the true value again in a relatively short time even when the identified parameter φ̂(k) identified in the sequential identification calculation greatly deviates from the true value. Accordingly, it is possible to effectively prevent the occurrence of such a problem that the state where the identified parameter φ̂(k) greatly deviates from the true value is maintained once the identified parameter φ̂(k) greatly deviates from the true value.
In each of
As shown in
Meanwhile, as shown in
In the embodiment described above, the current detecting part 301 corresponds to “current detecting means”, the voltage detecting part 302 to “voltage detecting means”, the state variable filter calculation part 304 to “identification calculation part” and “identification calculation means”, the adaptive identification calculation part 305 to “identification calculation part”, “identification calculation means”, and “initializing part”, and the open circuit voltage estimating part 306 to “open circuit voltage estimating part”, “limit processing part”, and “limit processing means”.
Although the embodiment of the present invention has been described above, this embodiment is described to facilitate the understandings of the present invention and is not described to limit the present invention. Accordingly, the elements disclosed in the embodiment described above are intended to include all of the design changes and equivalents pertaining to the technical scope of the present invention.
For example, in the embodiment described above, the following mode is given as an example. The determination of whether the “φn(k−1)≧δ2” is satisfied is performed for the highest-degree term parameter φn(k−1) of the previous processing (Step S4). When “φn (k−1)≧δ2” is satisfied, the sequential identification calculation is initialized (Step S5). Thereafter, the identification calculation processing for identifying the parameter φ̂(k) of the battery model is performed (Step S6). However, the present invention is not limited to this mode and the following mode may be employed. First, the identification calculation processing for identifying the parameter φ̂(k) of the battery model is performed and the determination of whether the “φn(k)≧δ2” is satisfied is then performed for the highest-degree term parameter φn(k) obtained as a result of the identification calculation processing. In this case, when it is determined that “φn(k)≧δ2” is satisfied, the sequential identification processing is initialized and the identification calculation processing for identifying the parameter φ̂(k) of the battery model is performed again.
This application claims priority from Japanese Patent Application No. 2010-162738 filed on Jul. 20, 2010 and Japanese Patent Application No. 2011-153665 filed on Jul. 12, 2011, the contents of which are incorporated herein by reference in its entirety.
The battery state estimating device and the battery state estimating method of the embodiment of the present invention estimate the terminal voltage of the secondary battery which is based on the predetermined battery model, by using the measured values of the current and the terminal voltage of the secondary battery, and sequentially identify the parameters of the battery model in such a way that the difference between the value based on the measured value of the terminal voltage and the estimate value of the terminal voltage converges to zero. When the certain parameter φn among the identified parameters is equal to or larger than the predetermined first threshold δ1, the value of the certain parameter φn is set to the first threshold δ1. This reduces the estimation error of the state inside the battery which is caused by the identification error of the certain parameter φn. The state inside the secondary battery can be thereby properly estimated. Accordingly, the battery state estimating device and the battery state estimating method of the embodiment of the present invention have industrial applicability.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2010-162738 | Jul 2010 | JP | national |
| 2011-153665 | Jul 2011 | JP | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/JP2011/066374 | 7/19/2011 | WO | 00 | 1/16/2013 |
