Example [parameter extraction]

In [7]:

import numpy as np
import pandas as pd
from pandas import DataFrame as df
from batterylearn.models import OCV, EcmCell
from batterylearn.utilities import ivp
from batterylearn.simulations import Data, Current, Learner,Simulator
import os
import matplotlib.pyplot as plt

import numpy as np import pandas as pd from pandas import DataFrame as df from batterylearn.models import OCV, EcmCell from batterylearn.utilities import ivp from batterylearn.simulations import Data, Current, Learner,Simulator import os import matplotlib.pyplot as plt

In [8]:

ocv = [
    2.725578,
    2.850545,
    2.952456,
    3.020007,
    3.072155,
    3.114243,
    3.149442,
    3.178211,
    3.196915,
    3.208162,
    3.21203,
    3.252616,
    3.284272,
    3.308415,
    3.311269,
    3.314391,
    3.336869,
    3.345692,
    3.351787,
    3.353211,
    3.354968,
    3.357175,
    3.360247,
    3.364758,
    3.372647,
    3.388452,
    3.421374,
    3.428617,
    3.433137,
    3.438826,
    3.444884,
    3.4514,
    3.458486,
    3.466429,
    3.475249,
    3.5,
]
soc = [
    0.397351,
    1,
    2,
    3,
    4,
    5,
    6,
    7,
    8,
    9,
    10,
    20,
    30,
    40,
    50,
    60,
    70,
    80,
    90,
    91,
    92,
    93,
    94,
    95,
    96,
    97,
    98,
    98.1,
    98.2,
    98.3,
    98.4,
    98.5,
    98.6,
    98.7,
    98.8,
    100.0,
]

c1 = OCV(name="curve1", ocv=ocv, soc=soc)

c1.display()

ocv = [ 2.725578, 2.850545, 2.952456, 3.020007, 3.072155, 3.114243, 3.149442, 3.178211, 3.196915, 3.208162, 3.21203, 3.252616, 3.284272, 3.308415, 3.311269, 3.314391, 3.336869, 3.345692, 3.351787, 3.353211, 3.354968, 3.357175, 3.360247, 3.364758, 3.372647, 3.388452, 3.421374, 3.428617, 3.433137, 3.438826, 3.444884, 3.4514, 3.458486, 3.466429, 3.475249, 3.5, ] soc = [ 0.397351, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 91, 92, 93, 94, 95, 96, 97, 98, 98.1, 98.2, 98.3, 98.4, 98.5, 98.6, 98.7, 98.8, 100.0, ] c1 = OCV(name="curve1", ocv=ocv, soc=soc) c1.display()

the training dataset has a gap , during which the cell is relaxing,

In [9]:

TESTDATA_FILEPATH = os.path.join(os.path.dirname(os.getcwd()),
'../tests/cycling/main-data-as-seriestocolumns-2022-08-02 23_32_07.xlsx')

# data come from https://web.calce.umd.edu/batteries/data.htm

schema = {
    "Time": "time",
    "Cell_LFP_CALCE_A1_007.max_current": "current",
    "Cell_LFP_CALCE_A1_007.max_v_tr": "vt",
    "rsv_i_dir": False,
}
d1 = Data(name="d1", df=None)
d1.fetch_file(TESTDATA_FILEPATH, schema=schema)
d1.to_abs_time()
d1.disp()

TESTDATA_FILEPATH = os.path.join(os.path.dirname(os.getcwd()), '../tests/cycling/main-data-as-seriestocolumns-2022-08-02 23_32_07.xlsx') # data come from https://web.calce.umd.edu/batteries/data.htm schema = { "Time": "time", "Cell_LFP_CALCE_A1_007.max_current": "current", "Cell_LFP_CALCE_A1_007.max_v_tr": "vt", "rsv_i_dir": False, } d1 = Data(name="d1", df=None) d1.fetch_file(TESTDATA_FILEPATH, schema=schema) d1.to_abs_time() d1.disp()

In [10]:

d1.df.interpolate(method='linear',inplace=True)
d1.df.dropna(inplace=True)
d1.disp()

d1.df.interpolate(method='linear',inplace=True) d1.df.dropna(inplace=True) d1.disp()

In [11]:

d1.df.reset_index(inplace=True)

d1.df.reset_index(inplace=True)

In [12]:

param_sim = {
    "R0": 0.1,
    "R1": 0.1,
    "tau1": 2,
    "R2": 0.1,
    "tau2": 1,
    "CAP": 1.1,
    "ce": 0.99,
    "v_limits": [1.5, 4.5],
    "SOC_RANGE": [0.0, 100.0],
}
m_sim = EcmCell(name="cell_model_sim", parameters=param_sim, curve=c1)
l2 = Learner(name="l2")
l2.attach(d1).attach(m_sim)
x0_sim = np.array([-0.05, -0.05, 100])

config = {
    "solver_type": "adaptive",
    "solution_name": "sol1",
    "max_step":np.inf,
    'maxiter':50,
    'maxfev':2000,
    'method':'Powell'}

method = "diff"
bounds=(
    (1e-4, 1),
    (1e-4, 1),
    (1, 10),
    (1e-4, 1),
    (.1, 10),
    (0.9, 1.2),
)

res = l2.fit_parameters(("cell_model_sim", "d1"), config, x0_sim, method,bounds)

param_sim = { "R0": 0.1, "R1": 0.1, "tau1": 2, "R2": 0.1, "tau2": 1, "CAP": 1.1, "ce": 0.99, "v_limits": [1.5, 4.5], "SOC_RANGE": [0.0, 100.0], } m_sim = EcmCell(name="cell_model_sim", parameters=param_sim, curve=c1) l2 = Learner(name="l2") l2.attach(d1).attach(m_sim) x0_sim = np.array([-0.05, -0.05, 100]) config = { "solver_type": "adaptive", "solution_name": "sol1", "max_step":np.inf, 'maxiter':50, 'maxfev':2000, 'method':'Powell'} method = "diff" bounds=( (1e-4, 1), (1e-4, 1), (1, 10), (1e-4, 1), (.1, 10), (0.9, 1.2), ) res = l2.fit_parameters(("cell_model_sim", "d1"), config, x0_sim, method,bounds)

/workspaces/batterylearn/.venv/lib/python3.10/site-packages/scipy/optimize/_differentialevolution.py:377: UserWarning: differential_evolution: the 'workers' keyword has overridden updating='immediate' to updating='deferred'
  with DifferentialEvolutionSolver(func, bounds, args=args,

differential_evolution step 1: f(x)= 0.00183315
differential_evolution step 2: f(x)= 0.00183315
differential_evolution step 3: f(x)= 0.00183315
differential_evolution step 4: f(x)= 0.00183315
differential_evolution step 5: f(x)= 0.00183315
differential_evolution step 6: f(x)= 0.00183315
differential_evolution step 7: f(x)= 0.00183315
differential_evolution step 8: f(x)= 0.00178681
differential_evolution step 9: f(x)= 0.00178681
differential_evolution step 10: f(x)= 0.00172918
differential_evolution step 11: f(x)= 0.00172918
differential_evolution step 12: f(x)= 0.00172918
differential_evolution step 13: f(x)= 0.00165522
differential_evolution step 14: f(x)= 0.00154616
differential_evolution step 15: f(x)= 0.00154616
differential_evolution step 16: f(x)= 0.00154616
differential_evolution step 17: f(x)= 0.00154616
differential_evolution step 18: f(x)= 0.00154616
differential_evolution step 19: f(x)= 0.00154616
differential_evolution step 20: f(x)= 0.00154616
differential_evolution step 21: f(x)= 0.00154616
differential_evolution step 22: f(x)= 0.00154616
differential_evolution step 23: f(x)= 0.00154616
differential_evolution step 24: f(x)= 0.00154616
differential_evolution step 25: f(x)= 0.00154616
differential_evolution step 26: f(x)= 0.00148526
differential_evolution step 27: f(x)= 0.00148526
differential_evolution step 28: f(x)= 0.00148526
differential_evolution step 29: f(x)= 0.00148526
differential_evolution step 30: f(x)= 0.00146498
differential_evolution step 31: f(x)= 0.00146498
differential_evolution step 32: f(x)= 0.00146498
differential_evolution step 33: f(x)= 0.00146498
differential_evolution step 34: f(x)= 0.00146498
differential_evolution step 35: f(x)= 0.00146498
differential_evolution step 36: f(x)= 0.00146498
differential_evolution step 37: f(x)= 0.00146498
differential_evolution step 38: f(x)= 0.00146498
differential_evolution step 39: f(x)= 0.00146498
differential_evolution step 40: f(x)= 0.00146498
differential_evolution step 41: f(x)= 0.00146498
differential_evolution step 42: f(x)= 0.00144569
differential_evolution step 43: f(x)= 0.00140692
differential_evolution step 44: f(x)= 0.00140692
differential_evolution step 45: f(x)= 0.00140692
differential_evolution step 46: f(x)= 0.00140692
differential_evolution step 47: f(x)= 0.00140692
differential_evolution step 48: f(x)= 0.00139927
differential_evolution step 49: f(x)= 0.00139927
differential_evolution step 50: f(x)= 0.00139927

In [13]:

a = list(map(lambda x,y:(y[1] - y[0])*x+y[0], res.x, bounds))

print('the extracted R0 is {}, R1 is {}, tau1 is {}, R2 is {}, tau2 is {}, CAP is {}'.format(*a))
print('tau1 is {} min'.format(a[2]))
print('tau2 is {} hours'.format(a[4]))

param_exr = {
    "R0": a[0],
    "R1": a[1],
    "tau1": a[2],
    "R2": a[3],
    "tau2": a[4],
    "CAP": a[5],
    "ce": 0.99,
    "v_limits": [1.5, 4.5],
    "SOC_RANGE": [0.0, 100.0],
}

m_exr = EcmCell(name="cell_model_exr", parameters=param_exr, curve=c1)

s3 = Simulator(name="s3")
s3.attach(d1).attach(m_exr)
d3 = s3.run(("cell_model_exr", "d1"), x0_sim)

fig,  ax = plt.subplots(5,1,sharex=True,figsize=(10,20))

# ax[0].set_ylim(bottom=2.5,top=3.8)
# ax[0].set_ylim(bottom=-.75,top=.1)
d1.df.plot(ax=ax[0],x='time',y='vt',label='exp_data')
d3.df.plot(ax=ax[0],x='time',y='vt',label='model' )
d1.df.plot(ax=ax[1],x='time',y='current',label='current')
d3.df.plot(ax=ax[2],x='time',y='u1',label='u1')
d3.df.plot(ax=ax[2],x='time',y='u2',label='u2')
d3.df.plot(ax=ax[3],x='time',y='ocv',label='ocv')
d3.df.plot(ax=ax[4],x='time',y='soc',label='soc')
plt.show()

a = list(map(lambda x,y:(y[1] - y[0])*x+y[0], res.x, bounds)) print('the extracted R0 is {}, R1 is {}, tau1 is {}, R2 is {}, tau2 is {}, CAP is {}'.format(*a)) print('tau1 is {} min'.format(a[2])) print('tau2 is {} hours'.format(a[4])) param_exr = { "R0": a[0], "R1": a[1], "tau1": a[2], "R2": a[3], "tau2": a[4], "CAP": a[5], "ce": 0.99, "v_limits": [1.5, 4.5], "SOC_RANGE": [0.0, 100.0], } m_exr = EcmCell(name="cell_model_exr", parameters=param_exr, curve=c1) s3 = Simulator(name="s3") s3.attach(d1).attach(m_exr) d3 = s3.run(("cell_model_exr", "d1"), x0_sim) fig, ax = plt.subplots(5,1,sharex=True,figsize=(10,20)) # ax[0].set_ylim(bottom=2.5,top=3.8) # ax[0].set_ylim(bottom=-.75,top=.1) d1.df.plot(ax=ax[0],x='time',y='vt',label='exp_data') d3.df.plot(ax=ax[0],x='time',y='vt',label='model' ) d1.df.plot(ax=ax[1],x='time',y='current',label='current') d3.df.plot(ax=ax[2],x='time',y='u1',label='u1') d3.df.plot(ax=ax[2],x='time',y='u2',label='u2') d3.df.plot(ax=ax[3],x='time',y='ocv',label='ocv') d3.df.plot(ax=ax[4],x='time',y='soc',label='soc') plt.show()

the extracted R0 is 0.0032225266289396294, R1 is 0.9314412701641623, tau1 is 9.90224655682317, R2 is 0.9559621406733357, tau2 is 0.3464159109798134, CAP is 1.0541511896094677
tau1 is 9.90224655682317 min
tau2 is 0.3464159109798134 hours