075582814553
The lithium-ion battery exhibits favorable characteristics, making it a suitable option for electric vehicles.
Because of a lithium-ion battery’s high energy density, low self-discharge rate, high voltage, long lifespan, high reliability, and quick recharging capabilities, it has drawn a lot of interest from both the scientific community and the automotive industry.
Compared to the various battery management systems, lithium-ion batteries are predominant in terms of specific energy and specific power. When you look at lithium-ion batteries and other electric vehicle batteries side by side in terms of nominal voltage, life cycle, depth of discharge, and efficiency, lithium-ion batteries seem to be the better choice for electric vehicle use.
Many applications, including those in the military, aerospace industry, electric vehicles, and portable devices, have benefited from the profitable qualities of lithium-ion batteries.
Although lithium-ion batteries have many appealing features, their pressurized, flammable liquid electrolyte can make them dangerous. Even with a comparatively tiny lithium-ion battery package, it can result in explosions and fires under the correct conditions.
For example, thermal runaways caused by overcharging lithium-ion batteries may result in leaks and explosions. However, overdischarging a lithium-ion battery can cause irreversible damage to the battery and hasten its aging process. The state-of-charge of the lithium-ion batteries must always be known in order to guarantee the safe charging and discharging functioning of lithium-ion batteries.
When the lithium-ion batteries are not in a safe operating condition, an accurate state-of-charge prediction allows the cut-off circuitry to disconnect them and initiate battery charging in a safe operating scenario.
The remaining quantity of charge in a battery is indicated by its state of charge, which is an essential parameter of batteries. An accurate state-of-charge calculation prolongs battery life and minimizes catastrophic battery failure.
Furthermore, a reliable and accurate state-of-charge estimate is essential to the effective functioning of electric vehicles. Battery age, ambient temperature, and a host of other variables make state-of-charge determination a challenging and intricate operation.
State-of-charge estimate techniques can be divided into five groups, as seen in Fig. 1. These groups include:
● Model-based estimation methods
● Data-driven estimation methods
● Look-up table approach,
● Coulomb counting method
● Hybrid method.
Every category uses a different methodology to assess state of charge performance. This article gives a quick idea of what model-based estimation methods and data-driven estimation methods are about.
The background process information is used in the design of the model-based state-of-charge estimate techniques, commonly referred to as white-box models. The conventional technique, or model-based method, is well-known for its ability to handle numerous issues, particularly in the engineering field.
To create strong rules that can effectively describe the behavior of the system, this approach frequently requires the practitioner to have a thorough knowledge of the system or process.
Because model-based state-of-charge estimate techniques rely on in-depth knowledge of the system, they can be quite potent and precise. A model-based approach is essential for solving many engineering and physics challenges.
For instance, it represents the earth's gravitational pull, the path of a projectile, and so on.
To achieve the ideal model of any system, there are theoretical and practical challenges.
In practical terms, it usually takes a lot of work, difficult experiments, and in-depth system research by domain specialists to design a reliable state-of-charge estimation model that might best characterize a system.
From a theoretical perspective, a depth of theoretical knowledge about the system is required for the model-based state-of-charge estimation approaches.
For example, scientists have combined different types of physics, like reaction polarization, ohmic polarization, solid-phase diffusion, liquid-phase diffusion, and open-circuit voltage, to make a simplified electrochemical model of lithium-ion batteries. Determining the battery parameters is challenging due to the numerous difficult mathematical equations involved in each of the physics of the proposed model.
In summary, inadequate past understanding of the system invariably results in poor model construction. To create a reliable model, domain experts must thus have a thorough understanding of the system's mechanical, electrical, electronic, chemical, and other elements.
However, the development of sizable data sets and potent computers has made it possible for a relatively new approach known as a data-driven, state-of-charge estimate. With little to no understanding of the underlying processes, data-driven approaches, also referred to as black-box models, are constructed using empirical findings.
The data-driven method does not require practitioners to have a thorough understanding of the background process because it heavily relies on data analysis from the process.
Using this method, a state-of-charge estimation model can be made even if the person does not know much about how batteries work or how chemicals react inside them. In this regard, modeling a complicated system using the data-driven approach takes less effort and expertise than using the model-based approach.
A long-short-term memory network can look at state-of-charge by only monitoring measurements of the battery, like current, voltage, and temperature; it doesn't need to know about the chemistry inside the battery, complex reactions, or estimating model parameters.
However, because data-driven methods depend so much on the data collected during the process, the quality of the data has a big impact on how well the model works and how accurate it is.
As an example, uneven data would lead to overfitting and underfitting, which are both types of bias in decision-making. These problems have been extensively studied, and researchers have come up with general guidelines for fixing them.
At its core, a data-driven method would only work well if a lot of relevant data were readily available. The data-driven method, on the other hand, wouldn't be very useful without these data.
A substantial body of research has been done on the application of model-based and data-driven estimation techniques in the field of state-of-charge estimation. In terms of state-of-charge estimates, both the model-based and data-driven techniques have produced noteworthy outcomes.
A model-based method is the best way to go from a statistical point of view if the model of the system is known ahead of time, according to the thorough review.
On the other hand, the data-driven method could work better than model-based solutions if the system is not well understood. To get the best of both worlds, several researchers have been attempting to combine the two methods.
Nevertheless, more research and development are moving in the direction of data-driven algorithm-based state-of-charge estimation due to the advancements in technology, including rapid processors, large data availability, and high-capacity storage devices.
● Accurate state-of-charge prediction is crucial for the safe charging and discharging of lithium-ion batteries and the effective functioning of electric vehicles.
● State-of-charge estimation methods can be divided into five groups, including model-based and data-driven estimation methods.
● Model-based methods require a thorough understanding of the system, while data-driven methods rely on data analysis from the process.
● Combining model-based and data-driven estimation methods can provide the best of both worlds.
● Advancements in technology, including rapid processors, large data availability, and high-capacity storage devices, are driving research and development towards data-driven algorithm-based state-of-charge estimation.
How, Dickson N. T., M. A. Hannan, M. S. Hossain Lipu, and Pin Jern Ker. “State of Charge Estimation for Lithium-Ion Batteries Using Model-Based and Data-Driven Methods: A Review.” IEEE Access 7 (2019): 136116–36. https://doi.org/10.1109/access.2019.2942213.
Catalogue
FPGA / CPLD
Memory
MOS 
MCU 
DSP
OCEP
Secondary 
Other 