摘要 |
The invention relates to a method for estimating the capacitance of a cell comprising: - automatically learning (102, 212) a function GSVM that returns the capacitance Cn,k3 based on the capacitance Cn,k3-1 at a previous time k3-1, and at least one physical characteristic cc1,k3 correlated with the change over time in the capacity of the cell and measured at time k3, wherein said automatic learning comprises the execution (108) of a supervised learning algorithm on a learning database, wherein said learning database contains capacitance values and, for each capacitance value, the value of the physical characteristic measured at the same time and the previous capacitance value; and - once the function GSVM is learned, at an earlier time k3+1, calculating (190) the capacitance prediction Cn,k3+1 using the state model defined by the following relation: Cn,k3+1 = GSVM(Cn,k3, cc1,k3+1). |