Cost-effective yet still precise ascertainment of the degradation state of a rechargeable battery
A method for ascertaining an approximation and/or a prognosis for the true degradation state of a rechargeable battery. The method includes: providing a time sequence of values of the degradation state ascertained using measuring technology for past points in time; providing a trained HMM, which indicates, as a function of the true degradation state, at which probability during the ascertainment using measuring technology which particular value of the degradation state is monitored, and at which probability the true degradation state is maintained for what length of time, and/or at which probability this true degradation state transitions to which worse degradation state in the next time step; from the monitored time sequence and the HMM, the most probable characteristic of the true degradation state in the past that is in agreement with the monitored time sequence is ascertained; the desired approximation and/or prognosis is evaluated based on the most probable characteristic.
1. A method for ascertaining an approximation and/or a prognosis for a true degradation state of a rechargeable battery of an at least partly electrically driven vehicle, the method comprising the following steps:
providing a time sequence, discretized into predefined time steps, of values of a degradation state ascertained using measuring technology, for points in time in the past;
providing a trained Hidden Markov Model (HMM), which indicates as a function of the true degradation state:
at which probability which particular value of the degradation state is monitored during the ascertainment using the measuring technology, and
at which probability the true degradation state is maintained for what length of time, and/or at which probability the true degradation state transitions to which worse degradation state in a next time step;
ascertaining, from the provided time sequence and the HMM, a most probable characteristic of the true degradation state in the past that is in agreement with the provided time sequence, wherein:
the approximation and/or prognosis is based on the most probable characteristic,
during discharging of the battery, time sequences of a clamping voltage, a discharge current, and a charge state of the battery are acquired, and the values of the degradation state are ascertained from the time sequences of the clamping voltage using a physical model of the degradation state, and
the physical model and the HMM are different from each other, wherein the most probable characteristic of the true degradation state in at least one past point in time that is in agreement with the provided time sequence is updated using a Viterbi algorithm, and wherein the approximation and/or prognosis is based on the most probable characteristic of the true degradation in the at least one past point in time that is in agreement with the provided time sequence.
2. The method as recited in claim 1 , wherein the values of the degradation state are ascertained based on a charge quantity that is output to the battery when charging the battery from a first clamping voltage to a second, higher clamping voltage.
3. The method as recited in claim 1 , wherein the values of the degradation state are ascertained based on a time period during which a clamping voltage of the battery drops from a first value to a second, lower value under loading by a predefined load.
4. The method as recited in claim 1 , wherein a time sequence of at least a temperature in the battery is acquired during the discharging and the temperature is taken into account in the physical model.
5. The method as recited in claim 1 , wherein when a prognosis ascertained for a future point in time for the true degradation state satisfies a predefined criterion, a time period up to the future point in time is considered a remaining usable service life of the battery.
6. The method as recited in claim 1 , wherein a traction battery of an at least partly electrically driven vehicle is selected as the battery.
7. A non-transitory machine-readable data carrier on which is stored a computer program for ascertaining an approximation and/or a prognosis for a true degradation state of a rechargeable battery of an least partly electrically driven vehicle, the computer program, when executed by one or more computers, causing the one or more computers to perform the following steps:
providing a time sequence, discretized into predefined time steps, of values of a degradation state ascertained using measuring technology, for points in time in the past;
providing a trained Hidden Markov Model (HMM), which indicates as a function of the true degradation state:
at which probability which particular value of the degradation state is monitored during the ascertainment using the measuring technology, and
at which probability the true degradation state is maintained for what length of time, and/or at which probability the true degradation state transitions to which worse degradation state in a next time step;
ascertaining, from the provided time sequence and the HMM, a most probable characteristic of the true degradation state in the past that is in agreement with the provided time sequence, wherein:
the approximation and/or prognosis is based on the most probable characteristic,
during discharging of the battery, time sequences of a clamping voltage, a discharge current, and a charge state of the battery are acquired, and the values of the degradation state are ascertained from the time sequences of the clamping voltage using a physical model of the degradation state, and
the physical model and the HMM are different from each other, wherein the most probable characteristic of the true degradation state in at least one past point in time that is in agreement with the provided time sequence is updated using a Viterbi algorithm, and wherein the approximation and/or prognosis is based on the most probable characteristic of the true degradation in the at least one past point in time that is in agreement with the provided time sequence.
8. A computer configured to ascertain an approximation and/or a prognosis for a true degradation state of a rechargeable battery, the computer configured to:
provide a time sequence, discretized into predefined time steps, of values of a degradation state ascertained using measuring technology, for points in time in the past;
provide a trained Hidden Markov Model (HMM), which indicates as a function of the true degradation state:
at which probability which particular value of the degradation state is monitored during the ascertainment using the measuring technology, and
at which probability the true degradation state is maintained for what length of time, and/or at which probability the true degradation state transitions to which worse degradation state in a next time step;
ascertain, from the provided time sequence and the HMM, a most probable characteristic of the true degradation state in the past that is in agreement with the provided time sequence, wherein:
a desired approximation and/or prognosis is based on the most probable characteristic,
during discharging of the battery, time sequences of a clamping voltage, a discharge current, and a charge state of the battery are acquired, and the values of the degradation state are ascertained from the time sequences of the clamping voltage using a physical model of the degradation state, and
the physical model and the HMM are different from each other, wherein the most probable characteristic of the true degradation state in at least one past point in time that is in agreement with the provided time sequence is updated using a Viterbi algorithm, and wherein the approximation and/or prognosis is based on the most probable characteristic of the true degradation in the at least one past point in time that is in agreement with the provided time sequence.