Battery Management Systems Using Artificial Intelligence

Hello guys, welcome back to our blog. Here in this article, we will discuss battery management systems using artificial intelligence, will artificial intelligence replace the algorithms such as fuzzy logic, Kalman filter, and any other algorithms used?, and will it become the future of BMS?. Well, let us explain to you guys.

If you have any electrical, electronics, and computer science doubts, then ask questions. You can also catch me on Instagram – CS Electrical & Electronics.

Also, read:

Battery Management Systems Using Artificial Intelligence

BMS is the heart of electric vehicles according to us because it will manage and protect the battery pack. There are various operations that BMS does such as cell balancing (it is a method to make the state of charge/voltage of each cell equal), state estimation (like a state of charge, state of health, state of power, state of energy, state of function), fault management (from faults such as over-voltage, under-voltage, over-current, over-temperature, under-temperature, and other external faults), checking charge level in the pack for efficient charging, and some other functions that battery management system will perform.

Read: Top 08 Functions Of The Battery Management System

Now, assume what will happen if all these functions are done using artificial intelligence. First, let me give a short intro to AI. Artificial intelligence means making the system think or behave like human intelligence or creating an artificial human brain. In AI we will train machines with some date sets to make systems more intelligent or behave like humans.

One thing is, to make battery management systems using artificial intelligence we need a huge amount of data set to train our algorithm. Once our algorithm is trained then it can predict the states, protects the pack from fault, and also can perform balancing operations. It has various advantages such as high accuracy and quick operation. The whole BMS will act as a human brain.

For example, if you want to replace the Kalman filter for the state of charge estimation using artificial intelligence, then first you have to generate a huge data set. You have to conduct various tests on the battery pack like a number of HPPC tests and then you need to train your AI algorithm with data set so that it can predict the state of charge. As the number of training data increases the accuracy of the system will increase. once the application is trained with huge data, it can predict SoC at different drive cycles.

The same thing you can use for fault management, the first thing is to collect a huge set of data by creating various faults under different conditions. Then use all these data to train your AI algorithm. Fault management means it will not only isolate your BMS but also manages the fault and once the system is back to normal condition, it will connect the BMS back to load. At present for controlling and managing, model-based and non-mondel-based algorithms are used.

ANN is one of the algorithms used for fault management. ANN is quite similar to artificial intelligence but ANN (artificial neural network) not only predicts the output from a dataset but also learn things by itself based on previous experience. For example, a child first time he doesn’t know about fire and he keeps his hands on fire, and next time he will not go near to fire because he knows it will hurt him based on previous experience same thing ANN algorithm will act. Not only it will predict the output from the data set but also from past experience.

If you want to implement ANN in your fault management, then first you have to train algorithms with different fault data, and later it will learn by itself and update the data set. When compare the artificial intelligence, the artificial neural network is more accurate and reliable. But the complexity of using ANN is high when compared to AI.

There are other algorithms used for state estimations like Kalman filter, fuzzy logic, coulomb counting, etc. these algorithms can be replaced with AI. Cell balancing is one of the important functions of BMS, here you can implement AI to control the switching of switches (MOSFET). Cell balancing means making the SoC/voltage of each cell equal. There are two types of cell balancing approach such as active cell balancing and passive cell balancing. In the case of active cell balancing, the SoC/voltage of each cell is made equal by transferring energy from one cell to another cell, or from one cell to a battery pack whereas, in passive cell balancing, the SoC/voltage of each cell is made equal by dissipating energy through resistors.

Read: Difference Between Active And Passive Cell Balancing

For cell balancing, proper switching of switches (MOSFETs) is very important, here you can use an AI algorithm to check the level of each cell, and based on it, switch on the MOSFET to dissipate energy via a resistor or to transfer energy from one cell to another cell.

Even you can replace the CCU (central control unit) and LCU (local control unit) algorithm with AI. BMS consists of one CCU (also known as the master) and a number of LCU (also known as slave). For each 12-14 cells, one LCU is connected to measure the cell level voltage, temperature, and other parameters, and then LCU will pass the signal to CCU. CCU is responsible for measuring pack level parameters such as voltage temperature, current, and states along with fault management and other safety protection operations handled by CCU.

All these above operations can be done by AI only thing is a proper huge amount of data set is required to train the whole BMS AI algorithm.

This was about “Battery Management Systems Using Artificial Intelligence“. I hope this article may help you all a lot. Thank you for reading.

Also, read:

About The Author

Chetan Shidling

Automotive | Technical Blogger | Engineer | Content Creator

See author's posts

Share Now