Predicting state of a battery
View Patent ↗An observation model for calculating a capacity maintenance ratio of a battery. The capacity maintenance ratio is calculated on the basis of basic data representing an internal state of the battery (such as a feature quantity extracted from a positive and negative AC impedance curve plot), and a transition model for mapping a temporally preceding internal state and deterioration environment (SOC, that is, the battery capacity that is actually available, temperature, etc.) onto a current internal state are prepared in advance. Using the above models, a transition path of an internal state having the maximum likelihood, that is, the minimum sum of squares of differences between predictive values calculated from the above models and an actual value, is determined using an optimization method that is preferably dynamic programming, and the capacity maintenance ratio is predicted.
1. A battery state prediction method for predicting a capacity maintenance ratio of a battery under control of a computer, the method comprising:
preparing an observation model for calculating said capacity maintenance ratio on the basis of basic data representing an internal state of said battery, and a transition model for mapping a temporally preceding internal state and deterioration environment onto a current internal state; and
determining said current internal state of said battery by finding out a transition path of said internal state having a minimum sum of squares of differences between predictive values calculated from said observation model and said transition model and an actual value, on the basis of a value of said battery measured at each time step, wherein said basic data representing said internal state of said battery comprises a feature quantity extracted from a positive and negative AC impedance curve plot and said observation model is based on a nonlinear regression model.
2. The method according to claim 1 , wherein finding out of said transition path of an internal state having said minimum sum of squares is performed using dynamic programming.
3. The battery state prediction method according to claim 1 , further comprising:
preparing data on future time-series deterioration environments of said battery; and
sequentially estimating said capacity maintenance ratio of said battery in the future by applying said observation model and said transition model to said determined current internal state and to said data on future time-series deterioration environments of said battery.
4. The method according to claim 1 , wherein said deterioration environment includes SOC and temperature.
5. A battery state prediction computer program product for predicting a capacity maintenance ratio of a battery under control of a computer, the computer program product comprising:
a computer readable storage medium having computer readable non-transient program code embodied therein, the computer readable program code comprising:
computer readable program code configured to perform the steps of: preparing an observation model for calculating said capacity maintenance ratio on the basis of basic data representing an internal state of said battery, and a transition model for mapping a temporally preceding internal state and deterioration environment onto a current internal state; and
determining said current internal state of said battery by finding out a transition path of said internal state having the minimum sum of squares of differences between predictive values calculated from said observation model and said transition model and an actual value, on the basis of a value of said battery measured at each time step, wherein said basic data representing said internal state of said battery comprises a feature quantity extracted from a positive and negative AC impedance curve plot and said observation model is based on a nonlinear regression model.
6. The computer program product according to claim 5 , wherein finding out said transition path of an internal state having said minimum sum of squares is performed using dynamic programming.
7. The computer program product according to claim 5 , wherein said computer program product further causes the computer to execute the steps of:
preparing data on future time-series deterioration environments of said battery; and
sequentially estimating said capacity maintenance ratio of said battery in the future by applying said observation model and said transition model to said determined current internal state and to said data on future time-series deterioration environments of said battery.
8. The computer program product according to claim 5 , wherein said deterioration environment includes SOC and temperature.