Gangho is a senior Mechanical Engineering student at Sungkyunkwan University(SKKU), Korea. He joined SKKU in 2018, and then he served in the Korean military from 2019 to 2021. In 2022, he worked as an intern in the data science department of Hyundai Mobis Co.,ltd, and he is currently participating in the West program by the Korean government from 08,2023. He has an interest in data science and robotics. In DSlab, he is working on applying the SINDy algorithm to microbiome data and seeking a certain part of research to develop a dynamical model from the behaviors of a car.

Title:  Using Time Constants of Li-Ion Batteries for Safety Evaluation

Authors: Mohsen Derakhshan, Damoon Soudbakhsh 

Abstract: Lithium-ion batteries (LiBs) are the preferred choice of energy storage in many aspects of modern life from cell phones to electric vehicles (EVs), because of many desired properties such as high energy and low self-discharge. However, they pose severe hazards if their safety is compromised such as after sustaining mechanical damage. Prior work on evaluating the safety of LiBs following substantial damage was not conclusive as no detectable voltage or capacity changes were observed in them. The current study proposes a powerful, efficient, and reliable tool that can provide a solution to this problem.

Therefore, we aim at quantifying the safety of LiBs following a mechanical load or impact. We created a method to detect mechanical damage to Li-ion cells from their electrical response. The method is based on measuring their impedance spectra, determining their distribution of relaxation times (DRT), and analyzing them. We Modeled a battery impedance based on the DRT by solving a ridge regression optimization that involves a series of inductors, resistors, and capacitors as passive electrical components. New criteria for determining the optimal value of ridge regression optimization is developed in a way that the number of output peaks in the DRT be connected to the physics of the system. We tested five cylindrical cells 18650 with graphite/LiFePO4, while four cylindrical batteries are used to study the effect of mechanical damage on Lithium-Ion battery time constants. In the mechanical damage experiment, two cells were used as controls, and two cells were subjected to the indentation experiment. Using a 12.7 mm hemispherical punch, two cells are indented 6.5mm before short-circuiting. The indentation was held after each 1 mm displacement to measure the impedance spectrum of the cells (in the last step, only 0.5 mm displacement is present). To account for the changes in the EIS measurements and decouple the effect of punch indentation from the order of the experiment and the OCV drops after each EIS measurement, we conducted similar EIS experiments on the control cells. Therefore, control cells were tested with the exact timing of the tests on two indented cells. One cylindrical cell is used to study the effect of temperature and SOC on the time constant of the cell’s internal processes. We measured the impedance spectra of LIBs utilizing an EIS instrument at different cell temperatures (-20 oC to +40 oC) and State-of-charge (0% and 100% SOC). Using the introduced approach results in 5-6 dominant peaks in the 0.01 Hz to 45 kHz range, with 4-5 peaks in the medium and low frequency and only one peak in the high-frequency part of the impedance. The number of dominant peaks agrees with the expected number of internal processes determined experimentally. We use the dependency of the peaks on temperature and SOC (State of Charge) to assign them to major processes (diffusion, charge transfer, SEI (Solid Electrolyte Interphase), and the changes in the properties of the electrodes and separator). Using our DRT formulation and criteria, we show that the indented cells have substantially different high-frequency characteristics than the control group (the changes of the height of high frequency peak is 2.5% for control cells and 36.0% for indented cells). This non-invasive method can detect hazardous mechanical damage to the EV batteries after a road crash or impact landings of drones. Other applications of the proposed approach include 1) EVs evaluation during standard crash tests, 2) planned impact and shock applications, and 3) regular safety checks.

Related Publications:
1- Derakhshan, M., Sahraei, E., & Soudbakhsh, D. (2022). Detecting mechanical indentation from the time constants of Li-ion batteries. Cell Reports Physical Science, 3(11).

2- Derakhshan, M., & Soudbakhsh, D. (2021). Temperature-dependent time constants of li-ion batteries. IEEE Control Systems Letters, 6, 2012-2017.

Renato is proposing his Ph.D. Thesis on Monday, October 9th, 2023.

Renato presented at the IMECE 2023, International Mechanical Engineering Congress & Exposition, ASME.

Title: Data-Driven Modeling for Accurate State-of-Charge Prediction of Li-Ion Batteries

Abstract: We present a physics-inspired input/output predictor of lithium-ion batteries (LiBs) for online state-of-charge (SOC) prediction. The complex electrochemical behavior of batteries results in nonlinear and high-dimensional dynamics. Accurate SOC prediction is paramount for increased performance, improved operational safety, and extended longevity of LiBs. The battery’s internal parameters are cell-dependent and change with operating conditions and battery health variations. We present a data-driven solution to discover governing equations pertaining to SOC dynamics from battery operando measurements. Our approach relaxes the need for detailed knowledge of the battery’s composition while maintaining prediction fidelity. The predictor consists of a library of candidate terms and a set of coefficients found via a sparsity-promoting algorithm. The library was enhanced with explicit physics-inspired terms to improve the predictor’s interpretability and generalizability. Further, we developed a Monte Carlo search of additional nonlinear terms to efficiently explore the high-dimensional search space and improved the characterization of highly nonlinear behaviors. Additionally, we developed a hyperparameter autotuning approach for identifying optimal coefficients that balance accuracy and complexity. The resulting SOC predictor achieved high predictive performance scores (RMSE) of 2.2 × 10-6 and 4.8 × 10-4, respectively, for training and validation on experimental results corresponding to a stochastic drive cycle. Furthermore, the predictor achieved an RMSE of 8.5 × 10-4 on unseen battery measurements corresponding to the standard US06 drive cycle, further showcasing the adaptability of the predictor and the enhanced modeling approach to new conditions.

Related Publications:

1. Rodriguez, R., Ahmadzadeh, O., Wang, Y., & Soudbakhsh, D. (2023). Data-driven Discovery of Lithium-Ion Battery State of Charge Dynamics. Journal of Dynamic Systems, Measurement, and Control, 1-13, https://doi.org/10.1115/1.4064026.
2. Rodriguez, R., Ahmadzadeh, O., Wang, Y., & Soudbakhsh, D. (2023, May). Discovering governing equations of li-ion batteries pertaining state of charge using input-output data. In 2023 American Control Conference (ACC) (pp. 3081-3086). IEEE, https://doi.org/10.23919/ACC55779.2023.10156114.

Renato presented at the ACC’23, American Control Conference, IEEE.

Title: Discovering Governing Equations of Li-ion Batteries Pertaining State of Charge Using Input-Output Data

Abstract: Lithium-ion batteries (LIBs) have complex electrochemical behaviors, which result in nonlinear and high-dimensional dynamics. The modeling of this complex system often requires models involving PDEs, which are costly to develop and require invasive experiments to identify battery parameters. Here, we propose a data-driven technique to discover nonlinear reduced-order models that govern state-of-charge (SOC) dynamics from non-invasive input/output data. Accurate SOC estimation is paramount for increased performance, improved operational safety, and extended longevity of LIBs. The SOC model is developed from a library of candidate terms via a sparsity-promoting algorithm and data generated by the Doyle-Fuller-Newman (P2D) model with a thermal model to characterize the cell’s thermal effects. We tuned the model’s performance and sparsity by exploring different combinations of candidate terms (basis functions) and data sampling rates. Using current, voltage, and SOC, the model was trained and validated on the UDDS city driving cycle. It achieved a predictive performance (RMSE) of 3e-5% and 0.22% for training and model validation, respectively. The generalizability of the model was assessed via cross-validation on the US06 highway driving cycle, where an RMSE of 0.47% was achieved. The modeling technique includes explicit physics-inspired terms, which allows for interpretable and generalizable models. Furthermore, the procedures and methods developed in this research are generic and can guide machine learning modeling of other dynamical systems.

Rodriguez, R., Ahmadzadeh, O., Wang, Y., & Soudbakhsh, D. (2023, May). Discovering governing equations of li-ion batteries pertaining state of charge using input-output data. In 2023 American Control Conference (ACC) (pp. 3081-3086). IEEE

https://doi.org/10.23919/ACC55779.2023.10156114.

Omidreza presented at the IMECE 2023, International Mechanical Engineering Congress & Exposition, ASME.

Title: Interpretable Machine Learning Modeling of Li-ion Batteries

Abstract: Lithium-ion batteries (LIBs) are present in many modern applications due to several desirable properties such as high energy and power densities. Accurate real-time modeling of LIBs improves their operation and safety. Traditionally, equivalent circuit models (ECM) have been used to model LIBs due to their simplicity. These models utilize passive electrical components such as resistors and capacitors to model the battery’s responses. However, their lack of connection to physics results in poor extrapolation performance, and they require limiting the operating range and life of LIBs. These drawbacks have led to an increase in the popularity of physics-based models of LIBs in real-time applications over the past few years. However, developing these models is very cumbersome and requires several measurements and information that often are unavailable and change with the operational conditions and life of the batteries. We propose a novel solution for battery management systems through interpretable machine learning (ML) modeling. We identify the governing equations of LIBs without requiring the in-situ measurements and proprietary information needed by physics-based models. To address the common issue of overfitting and finding a wrong fit with many ML techniques, which results in poor performance in unseen scenarios, we propose a novel physics-informed reduced-order nonlinear model of LIBs. The model’s input is the electrical current and the measurable output is the voltage. We seek an input/output based formulation that predict the dynamics of LiBs., and SOC as the output of the ML model. We used Sequentially Thresholded Ridge regression (STRidge) to promote the model’s sparsity. The technique includes physics-based functions and employs sparse regression to balance the accuracy and complexity of the model using measured data. These terms were associated with the solution of the Doyle-Fuller-Newman (DFN) model, which is a mechanistic model. The added observables (functions of the input/output data) are solid and electrolyte concentration, the Butler-Volmer equation, and solid and electrolyte electric potentials. Furthermore, we improve our ML models by i) augmenting the method to determine the battery’s governing equations with noisy measurements accurately, and ii) restructuring the optimization problem and introducing the sparsifying parameters as hyperparameters and tuning them using a training and a validation dataset, hence resulting in a more generalizable model. Sparsifying parameters are thresholds and regularization parameters that nullify the less important terms and adjust the coefficient values, respectively to balance model accuracy and complexity. We use training data with known input/output to find sparse models for different ranges of hyperparameters. The identified sparse model for each hyperparameter is assessed with the validation data set whose input is only known. The hyperparameters that provide the lowest cost function for the validation data are selected. In addition, we present a robust modeling technique for noisy measurements. This model uses a Kalman filter approach, where the sparse terms are updated based on the Kalman filter to remove the noise from the voltage data and enhance the state of charge (SOC) prediction. We have tested the method using data from advanced chemistry that is used to develop new electric vehicle applications (21700 cylindrical cells, NMC811). We used uniformly distributed electrical current signals up to 2C/4C charge/discharge rates for training the model and the US highway profile (US06) for the validation set. We showed the model’s accuracy using the Urban Dynamometer Driving Schedule (UDDS) as the unseen test data. The model predicted the response with less than 8.3×10^−5 normalized root mean square error (NRMSE) for SOC and Voltage predictions.

Related Publications:

1. O. Ahmadzadeh, R. Rodriguez, Y. Wang, D. Soudbakhsh, A physics-inspired machine learning nonlinear model of li-ion batteries, in: 2023 American Control Conference (ACC), IEEE, 2023, pp. 3087–3092. 10.23919/ACC55779.2023.10156368
2. O. Ahmadzadeh, Y. Wang, D. Soudbakhsh, Sparse modeling of energy storage systems in presence of noise, IFAC-PapersOnLine 56 (2) (2023) 3764–3769. https://doi.org/10.1016/j.ifacol.2023.10.1546

Title: Effect of Temperature and SOC on Frequency-Response Modeling of Li-ion Batteries

Abstract: This paper investigates the choice of equivalent circuit models (ECMs) for modeling Li-ion batteries (LiBs). ECMs with distributed elements (dECMs) are the most common tools to analyze LiB impedance spectra. However, the choice of the model is ambiguous. In this work, we investigate the validity of 14 typical models to accurately represent the impedance spectra of the cells under various operating conditions such as cycling, various State-of-Charge, and temperatures. These models represent a more extensive set as such ECMs are degenerate, and several models are equivalent.
The model parameters were fit to the experimental data using a complex nonlinear least squares (CNLS) approach.
We used 7 Li-ion cylindrical cells with graphite/LiFePO_4 material in our studies and measured electrochemical impedance spectroscopy (EIS) under several scenarios: i) new cells, ii) after cycling, iii) resting at different temperatures, and iv) indented cells.
The models with the worst fit were excluded from the study after each fitting set. Our study showed that two of the 14 original models are general enough to represent the tested scenarios without overfitting; therefore, these models can be used to analyze the effect of temperature, SOC, and mechanical loading.

DOI: 10.1109/CCTA54093.2023.10253177

Derakhshan, M., Shankin, H. M., Lohan, L. A., & Soudbakhsh, D. (2023, August). Effect of Temperature and SOC on Frequency-Response Modeling of Li-ion Batteries. In 2023 IEEE Conference on Control Technology and Applications (CCTA) (pp. 388-393). IEEE.