A Project Report On

STATE OF POWER PREDICTION OF

LITHIUM-ION BATTERY IN ELECTRIC VEHICLE

By MARISERLA PAVAN KUMAR

[email protected]

Under Guidance Of
Dr. ARGHYA MITRA
Department of Electrical Engineering
Visvesvaraya National Institute of Technology, Nagpur
MOHD. FASIL (Senior Engineer)
B. SUJEESH (Engineer)
TATA ELXSI

1
Contents
1. INTRODUCTION 3

2. LITERATURE REVIEW 4

3. PROBLEM STATEMENT 8

4. METHODOLOGY 8

5. RESULT ANALYSIS 16

6. CONCLUSION 17

7. PLAN FOR NEXT SIX MONTHS 17

8. REFERENCES 19

2
INTRODUCTION:
State of Power (SoP):
A battery’s state-of-power (SOP) is defined as the ratio of peak
power to nominal power. The peak power, based on present
battery-pack conditions, is the maximum power that may be
maintained constant for T seconds without violating preset
operational design limits on battery voltage, SOC, power, or
current.

This indicator is very important to ensure that the charge or

discharge power does not exceed certain limits with the aim of
using the battery as good as possible to extend its life
expectancy. Also, in peak power applications this indicator can
turn useful to define conditions in the battery to be able to make
big charges or discharges.
The state of power depends highly on the state of charge, the
capacity of the battery and its initial features, chemistry and
battery voltage so it is obtained in a second step of a battery
study.

3
LITERATURE SURVEY:
The state of power (SOP) of a battery system reflects the
maximum available power to be output or input in a short
period. SOP is the primary reference index for EVs to realize the
functions of accelerating, feedback braking and peak cutting and
valley filling, which are crucial for battery energy management.
Most existing methods for online SOP estimation employ an
equivalent-circuit model (ECM) to simulate battery dynamic
behavior in a prediction window. Assumption is made such that
batteries operating at one of the three operation modes namely
constant current (CC), Constant voltage (CV), Constant current
and Constant voltage (CCCV).
Many efforts have been made in the past decades for the
improved performance of online SOP estimation from three
aspects: model structure, online parameter identification
technique, SOP estimation algorithm.
[1] First, the online, model-based SOC estimation is fulfilled by
modified moving horizon estimation (mMHE) for better
convergence and fault tolerance. Second, the model parameters
are periodically updated by virtue of the mMHE -type
optimization with a relatively long horizon. Third, the ampere
hour integral and the estimated SOC are employed to realize the
capacity estimation offline. Given updated states and parameters,
the model-based real-time SOP estimation reliably predicts the

4
battery peak power respecting multiple operational constraints.
Finally, the effectiveness and resilience of the joint SOC/SOH/SOP
estimation is demonstrated through a number of experiments.
Considering the underlying coupling and characteristics among
these states, a multi-time-scale estimation framework is
developed for accurate estimates and moderate computational
cost.
[2] To solve the problem of, a system-model-based SOP
prediction method is first proposed in this work. Specifically,
based on the formulated system model and generic state-space
representation, the challenge of nonmonotonic system state
evolution, arising from the dynamic parallel current distribution,
is identified and systematically addressed by the proposed
method. As demonstrated by tests on a battery system set up
with experimentally verified parameter values, the proposed
method outperforms the commonly applied cell-SOP-based
methods for providing a more accurate and reliable prediction of
the battery system SOP. Proposed and derived model results is
verified through experimental tests on EV.
[3] This paper proposes a novel polarization voltage (NPV) model
based on current and time by simplifying the equation describing
the polarization characteristics in the electrochemical model.
The NPV model is simple and implementable, and the
parameters can be identified by using just small-batch data, and
it is timesaving for practical testing with simple structures. The
Li+ concentration is directly proportional to the current and

5
time, so the polarization overpotential is also logarithmic to the
current and time. The polarization voltage will be affected by the
resistance Rohm and increase with the increase of the current.
NPV model based on current and time. This model is simple and
implementable to the high-precision prediction the voltage and
SOP at t-second (t > 0).
[4] First, a linearized equivalent circuit-based battery model,
together with an affine projection algorithm is used to estimate
the model parameters. Then the state of health (SOH) estimator
is triggered weekly or semi-monthly offline to update capacity
based on the three-dimensional response surface open circuit
voltage model and particle swarm optimization algorithm for
accurate online SOC and state of power (SOP) estimation. At last,
the Unscented Kalman Filter utilizes the estimated model
parameters and updated capacity to estimate SOC online and
the SOP estimator provides the power limitations considering
SOC, current and voltage constraints, taking advantage of the
information from both SOH and SOC estimators. Experiments
show that the relative error of the SOH estimator is under 1% in
all aging states whatever the loading profile is. The mean
absolute SOC estimation error is under 1.6%. The SOP estimator
is validated by means of the calibrated battery model based on
the HPPC test and its performance is ideal.

6
[5] To address the issue of neglecting correlation while
estimating a state of system authors suggested joint estimation
of states. A machine learning method called LSTM network is
proposed to estimate battery SOC. A multiple-constraints
method is used to estimate battery SOP, which can provide an
effective reference for drone flight strategy. In this paper, joint
state estimation of SOC and SOP is realized. According to the
coupling relationship between two parameters, the result of SOC
estimation can be used as a parameter of SOP estimation.
The LSTM method with KF can be used to estimate SOC more
accurately, SOP estimation, multi-constraint estimation is used.
Compared with HPPC method, this method is more accurate.
[6] This is a review paper suggesting different techniques for SOP
prediction. Broadly classified into 2 types namely:
(i)Based on Adaptive characteristic maps.
(ii)Based on equivalent circuit models.
Comparison among different methods is done based on factors
like Computation time, Analytical complexity and accuracy etc.
Current dependence of charge transfer resistance is also
discussed for all methods.
According to current state of art it is summarized that techniques
based on equivalent circuit models are dominating techniques
for SOAP prediction.

7
PROBLEM STATEMENT:
Prediction of SOP considering SOE as input for addressing range
problem under critical power application.

METHODOLOGY
Input parameters for SOP prediction

Fig (1)-Skeleton diagram for SOP prediction

8
WORK DONE SO FAR..

SIMULATION CODE FOR BATTERY EQUIVALENT CIRCUIT

PARAMETER ESTIMATION:
% Using MathWorks tools, estimation techniques, and measured lithium-
ion
% or lead acid battery data, you can generate parameters for the
% Equivalent Circuit Battery block. The Equivalent Circuit Battery
% block implements a resistor-capacitor (RC) circuit battery with open
% circuit voltage, series resistance, and 1 through N RC pairs. The
% number of RC pairs reflects the number of time constants that
% characterize the battery transients. Typically, the number of RC pairs
% ranges from 1 through 5.

%% Step 1: Load and Preprocess Data

%
% Create a Pulse Sequence object, which represents a pulse sequence
% experiment
psObj = Battery.PulseSequence;
disp(psObj)

% Load in data
% Specify the file name
FileName = 'Synthetic_LiPo_PulseDischarge.mat';

% Read the raw data

[time,voltage,current] = Battery.loadDataFromMatFile(FileName);

% Add the data to the PulseSequence

psObj.addData(time,voltage,current);

% Review the data

psObj.plot();

9
% Break up the data into Battery.Pulse objects
%
% Create the Pulse objects within the PulseSequence. This creates a
% parameter object that contains look-up tables of the correct size, given
% the number of pulses
psObj.createPulses(...
'CurrentOnThreshold',0.1,... %minimum current magnitude to identify
pulse events
'NumRCBranches',3,... %how many RC pairs in the model
'RCBranchesUse2TimeConstants',false,... %do RC pairs have different
time constant for discharge and rest?
'PreBufferSamples',10,... %how many samples to include before the
current pulse starts
'PostBufferSamples',15); %how many samples to include after the next
pulse starts

% Specify the Simulink model that matches the number of RC branches

and
% time constants:
psObj.ModelName = 'BatteryEstim3RC_PTBS';
% Plot the identified pulses in the data
psObj.plotIdentifiedPulses();

%Note: if for some reason we want to exclude some of the pulses, you can
do
%something like this:
% psObj.removePulses(indexToRemove);
% psObj.plotIdentifiedPulses();

%% Step 2: Determine the Number of RC Pairs

%
% This step helps us decide how many RC pairs should be used in the
model.
% More RC pairs add complexity and might over-fit the data. Too few
% increases the fit error. Note: If you decide to change the number of
% pairs, you need to rerun createPulses above and change

10
NumRCBranches to
% the new value.

% Pick a pulse near the beginning, middle, and end. Note: you could run all
% of them if you want.
PulsesToTest = [1 floor(psObj.NumPulses/2), psObj.NumPulses-1];
% Perform the comparison
psObj.Pulse(PulsesToTest).compareRelaxationTau();

%% Step 3: Estimate Parameters

% Set settings
% Pull out the parameters. Only update parameters once because it
changes
% the history each time they update.
Params = psObj.Parameters;

% Set Em constraints and initial guesses (or don't and try the defaults)
% Params.Em(:) = 4.1;
% Params.EmMin(:) = 3.4;
% Params.EmMax(:) = 4.2;
% Set R0 constraints and initial guesses (or don't and try the defaults)
% Params.R0(:) = 0.001;
% Params.R0Min(:) = 0.0001;
% Params.R0Max(:) = 0.1;

% Set Tx constraints and initial guesses. It is important to default each

% time constant Tx to different values, otherwise the optimizer may not
% pull them apart.
Params.Tx(1,:,:) = 5;
Params.Tx(2,:,:) = 50;
Params.Tx(3,:,:) = 200;

Params.TxMin(1,:,:) = 1;
Params.TxMax(1,:,:) = 50;
Params.TxMin(2,:,:) = 20;
Params.TxMax(2,:,:) = 1000;

11
Params.TxMin(3,:,:) = 100;
Params.TxMax(3,:,:) = 3500; %don't set this bigger than the relaxation time
available
% Set Rx constraints and initial guesses (or don't and try the defaults)
% Params.Rx(:) = 0.01;
% Params.RxMin(:) = 0.0001;
% Params.RxMax(:) = 0.5;
% Update parameters
psObj.Parameters = Params;

% Estimate initial Em and R0 values

%
% This step inspects the voltage immediately before and after the current
% is applied and removed at the start and end of each pulse. It uses that
% for a raw calculation estimating what the open-circuit voltage (Em) and
% the series resistance R0 should be.

psObj.estimateInitialEmR0(...
'SetEmConstraints',false,... %Update EmMin or EmMax values based on
what we learn here
'EstimateEm',true,... %Keep this on to perform Em estimates
'EstimateR0',true); %Keep this on to perform R0 estimates

% Plot results
psObj.plotLatestParameters();

% Get initial Tx (Tau) values

%
% This step performs curve fitting on the pulse relaxation to estimate the
% RC time constant at each SOC.
psObj.estimateInitialTau(...
'UpdateEndingEm',false,... %Keep this on to update Em estimates at the
end of relaxations, based on the curve fit
'ShowPlots',true,... %Set this true if you want to see plots while this runs
'ReusePlotFigure',true,... %Set this true to overwrite the plots in the
same figure
'UseLoadData',false,... %Set this true if you want to estimate Time
12
constants from the load part of the pulse, instead of relaxation
'PlotDelay',0.5); %Set this to add delay so you can see the plots

% Plot results
psObj.plotLatestParameters(); %See what the parameters look like so far
psObj.plotSimulationResults(); %See what the result looks like so far

%
% Get initial Em and Rx values using a linear system approach - pulse by
% pulse
%
% This step takes the data for each pulse and treats it as a linear system
% It attempts to fit the Rx values for each RC branch. Optionally, you can
% allow it to adjust the Em and R0 values, and if these are adjusted, you
% also have the option whether to retain the optimized values of these or
% to discard them.

psObj.estimateInitialEmRx(...
'IgnoreRelaxation',false,... %Set this true if you want to ignore the
relaxation periods during this step
'ShowPlots',true,... %Set this true if you want to see plots while this runs
'ShowBeforePlots',true,... %Set this true if you want to see the 'before'
value on the plots
'PlotDelay',0.5,... %Set this to add delay so you can see the plots
'EstimateEm',true,... %Set this true to allow the optimizer to change Em
further in this step
'RetainEm',true,... %Set this true keep any changes made to Em in this
step
'EstimateR0',true,... %Set this true to allow the optimizer to change R0
further in this step
'RetainR0',true); %Set this true keep any changes made to R0 in this
step

% Plot results
psObj.plotLatestParameters(); %See what the parameters look like so far
psObj.plotSimulationResults(); %See what the result looks like so far

13
% Perform SDO Estimation
SDOOptimizeOptions = sdo.OptimizeOptions(...
'OptimizedModel',psObj.ModelName,...
'Method','lsqnonlin',...
'UseParallel','always');

psObj.estimateParameters(...
'CarryParamToNextPulse',true,... %Set this true to use the final
parameter values from the prior pulse and SOC as initial values for the next
pulse and SOC
'SDOOptimizeOptions',SDOOptimizeOptions,... %Specify the SDO
options object
'ShowPlots',true,... %Set this true if you want to see plots while this runs
'EstimateEm',true,... %Set this true to allow the optimizer to change Em
further in this step
'RetainEm',true,... %Set this true keep any changes made to Em in this
step
'EstimateR0',true,... %Set this true to allow the optimizer to change R0
further in this step
'RetainR0',true); %Set this true keep any changes made to R0 in this
step

% Plot results
psObj.plotLatestParameters(); %See what the parameters look like so far
psObj.plotSimulationResults(); %See what the result looks like so far

%% Step 4: Set Equivalent Circuit Battery Block Parameters

%
% The experiment was run at ambient temperature (303°K) only. Repeat
% the tables across the operating temperature range. If the discharge
% experiment was run at 2 different constant temperatures, then include
% these in the tables below.
EmPrime = repmat(Em,2,1)';
R0Prime = repmat(R0,2,1)';
14
SOC_LUTPrime = SOC_LUT;
TempPrime = [303 315.15];
CapacityAhPrime = [CapacityAh CapacityAh];

R1Prime = repmat(Rx(1,:),2,1)';
C1Prime = repmat(Tx(1,:)./Rx(1,:),2,1)';
R2Prime = repmat(Rx(2,:),2,1)';
C2Prime = repmat(Tx(2,:)./Rx(2,:),2,1)';
R3Prime = repmat(Rx(3,:),2,1)';
C3Prime = repmat(Tx(3,:)./Rx(3,:),2,1)';
open_system('BatteryEstim3RC_PTBS_EQ');

-----------------------------END OF THE CODE------------------------------

15
SIMULATION:

Fig (3)-CELL EQUIVALENT SIMULATION MODEL FOR PARAMETER ESTIMATION

Fig (4)- DETAILED EQUIVALENT CELL PARAMETER ESTIMATION MODEL

16
RESULTS

Fig (5)- VOLTAGE AND SOC DURING DISCHARGE PERIOD

17
 Fig (6)- PARAMETER VARIATION WITH RESPECT TO SOC

CONCLUSION:
Hence, lithium-ion battery equivalent cell model parameters are
estimated.

PLAN FOR NEXT SIX MONTHS:

Conducting HPPC test for real time static parameter estimation
and implementing SOP algorithm which includes SOE constraint.

18
REFERENCES:
[1] Xiaosong Hua .; Haifu Jiangb .; Fei Fenga .; Bo Liuc . “An
enhanced multi-state estimation hierarchy for advanced lithium-
ion battery management”. ELSEVIER journal-Applied Energy 257
(2020) 114019.
[2] Weiji Han , Faisal Altaf , Changfu Zou , Torsten Wik . “State of
Power Prediction for Battery Systems With Parallel-Connected
Units”. IEEE TRANSACTIONS ON TRANSPORTATION
ELECTRIFICATION, VOL. 8, NO. 1, MARCH 2022.
[3] Lin, P.; Wang, Z.; Jin, P.; Hong, J., Novel Polarization Voltage
Model: Accurate Voltage and State of Power Prediction. IEEE
Access 2020, 8, 92039–92049.
[4] Zhang, T.; Guo, N.; Sun, X.; Fan, J.; Yang, N.; Song, J. A
Systematic Framework for State of Charge, State of Health and
State of Power Co-Estimation of Lithium-Ion Battery in Electric
Vehicles. Sustainability. 2021, 13, 5166.
[5] Cheng, W.; Yi, Z.; Liang, J.; Song, Y.; Liu, D. An SOC and SOP
Joint Estimation Method of Lithium-ion Batteries in Unmanned
Aerial Vehicles. In Proceedings of the 2020 International
Conference on Sensing, Measurement & Data Analytics in the
Era of Artificial Intelligence (ICSMD), Xi'an, China, 5–17 October
2020; pp. 247–252.
[6] A. Farmanna , D.U. Sauer, A comprehensive review of on-
board State-of-Available-Power prediction techniques for
19
lithium-ion batteries in electric vehicles, J Power Sources 329
(2016) 129–137.
[7] Battery management systems vol-II, Equivalent circuit
methods, George L.Plett.
[8] Battery management systems vol-I, Battery Modelling,
George L.Plett.

20