Autonomous screening and optimization of battery formation and cycling procedures

Patent metadata

Field	Value
Publication number	US-2019115778-A1
Application number	US-201816161790-A
Country	US
Kind code	A1
Filing date	Oct 16, 2018
Priority date	Oct 17, 2017
Publication date	Apr 18, 2019
Grant date	—

Abstract

A method of probing a multidimensional parameter space of battery cell test protocols is provided that includes defining a parameter space for a plurality of battery cells under test, discretizing the parameter space, collecting a preliminary set of cells being cycled to failure for sampling policies from across the parameter space and include multiple repetitions of the policy, specifying resource hyperparameters, parameter space hyperparameters, and algorithm hyperparameters, selecting a random subset of charging policies, testing the random subset of charging policies until a number of cycles required for early prediction of battery lifetime is achieved, inputting cycle data for early prediction into an early prediction algorithm to obtain early predictions, inputting the early predictions into an optimal experimental design (OED) algorithm to obtain recommendations for running at least one next test, running the recommended tests by repeating from the random subset testing step above, and validating final recommended policies.

First claim

1 ) A method of probing a multidimensional parameter space of battery cell formation and cycling protocols, comprising:
a) defining a parameter space for a plurality of battery cells being optimized; b) specifying hyperparameters, wherein said hyperparameters comprise resource hyperparameters, parameter space hyperparameters, and algorithm hyperparameters; c) selecting a subset of said charging policies, including repetitions of policies; d) testing said subset of said policies, using a battery cycling instrument, until a number of cycles required for accuracy achieved; e) employing an optimal experimental design (OED) algorithm to obtain recommendations for running at least one next test; f) running said recommended tests by repeating f)-i) above; and g) validating final recommended policies.

2 ) The method according to claim 1 , wherein said parameter space comprises a number of cycling steps, a cycling time, a state-of-charge (SOC) range, and a boundary on a minimum and maximum current, voltage, resistance and temperature, or temperature, per said cycling step.

3 ) The method according to claim 1 , wherein said parameter space comprises a multi-step parameter space to optimize formation cycling or charging rate in a series of defined ranges of said SOC within a specified amount of time, wherein each said step controls a percentage of each said SOC range, wherein each said SOC range is independent from the other said SOC ranges, wherein a final said SOC range is a summation of all said SOC ranges prior to said final SOC range.

4 ) The method according to claim 1 , wherein said resource hyperparameters comprise a number of available testing channels, and a number of batches.

5 ) The method according to claim 1 , wherein said parameter space hyperparameters comprise a mean and standard deviation of a lifetime across all said policies, and a standard deviation of a single said policy tested multiple times.

6 ) The method according to claim 1 , wherein said algorithm hyperparameters comprise a degree of similarity between neighboring said policies in said parameter space, an exploration constant to control a balance of exploration versus exploitation, and a decay constant of said exploitation constant per round.

7 ) The method according to claim 1 , wherein said preliminary set of cells are configured to generate data to develop an early prediction model, quantify a mean, a standard deviation, and a range of lifetime over said parameter space, and quantify an intrinsic cell-to-cell variation for nominally identical cells cycled with nominally identical cycling conditions.

8 ) The method according to claim 1 further comprising a multi-phase said OED, wherein said multi-phase OED comprises a first round and a second round of closed-loop testing, wherein said first round comprises performing a preliminary classification of policies into a low-lifetime policy group or a high-lifetime policy group, wherein quantitative prediction is not required, wherein said second round comprises implementing said a)-j) above.

9 ) The method according to claim 1 further comprising a dynamic early prediction, wherein said dynamic early prediction comprises said collected preliminary set of cells is relaxed in size as more data is collected if a confidence in said prediction is increased.

10 ) The method according to claim 1 further comprising multi-cell sampling per policy within a test policy round, wherein said multi-cell sampling is directed to one or more cells of interest within said test policy round.

Assignees

• Univ Leland Stanford Junior

Inventors

• Ermon Stefano
• Chueh William C
• Grover Aditya
• Markov Todor Mihaylov
• Perkins Nicholas
• Attia Peter M

Classifications

• G06F30/20 Primary

  Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) · CPC title

• G06F2119/04

  Ageing analysis or optimisation against ageing · CPC title

• G05B13/048

  using a predictor · CPC title

• G05B13/042

  in which a parameter or coefficient is automatically adjusted to optimise the performance · CPC title

• H02J2103/30

  Simulating, planning, modelling, reliability check or computer assisted design [CAD] of electric power networks · CPC title