EV adoption dependent on AI-based solutions

In order to achieve net zero carbon emissions by 2050, and in aligning with the Paris Agreement’s goal of limiting global warming to less than 1.5 degree Celsius, many countries have started to impose a target of phasing out fossil fuel vehicles.

In this context, electric vehicles (EVs) are seen as the alternative to replace fossil fuel vehicles.

To fully exploit the benefits of EVs, they must be powered by renewable energy, which will increase the share of renewable energy in the energy mix (Sustainable Development Goal – target 7.2) and will minimise air pollution and related health impacts (Sustainable Development Goal – target 3.9).

The number of battery EVs in the world has grown exponentially over the last decade, with a total of 18 million battery EVs by the end of 2022, compared to 1.2 million in 2016. However, one of the main challenges of EVs is the reliability of the battery unit.

Battery degradation typically starts as soon as they are manufactured, and they must be replaced for safe operation when 70-80 per cent of their initial capacity is still present.

Batteries that exceed this threshold are more likely to malfunction, leading to serious economic losses and safety risks. To operate EVs safely and provide prior notification of potential battery failure, it is essential to develop a system that correctly calculates the battery’s remaining useful life (RUL) and capacity.

To overcome these issues, researchers from Universiti Tenaga Nasional (Uniten), Malaysia, and Universitas Pertamina, Indonesia, have taken the challenge to contribute to the development of an intelligent system to predict the RUL and capacity of batteries for EVs. Through an international joint research programme, the project undertaken by both parties aims at developing a solution that involves optimised RUL and future capacity estimation methods by using artificial intelligence (AI)-based algorithms.

In the current market, EVs use lithium-based batteries to store the electrical charges needed to power them. The methods to predict the RUL and capacity of lithium-ion batteries can be broadly divided into three categories: data-driven methods, model-based methods, and a hybrid of the data-driven and model-based approaches.

The model-based approach may provide a thorough electrochemical understanding of the battery ageing process and is suited for practically all conditions and operating modes. However, this approach requires an advanced filtering technique that leads to high computational complexity, and the uncertainty of the parameters of the battery in the training model was not considered.

On the other hand, the data-driven approach predicts RUL and battery cycle capacity without the need for extensive physical modelling, by examining the key components extracted by machine learning algorithms based on the measured degradation data. The drawback of this method is that the determination of the network parameters needs to be done through a time-consuming and labour-intensive trial-and-error process.

Similar to the model-based method, the reported studies for the data-driven approach did not consider the uncertainty of the battery’s parameters. Therefore, the main aim of this international joint research project is to tackle the problems by developing an improved method for predicting the future capacity and RUL of batteries for EVs.

To achieve this goal, researchers from Uniten and Universitas Pertamina will employ an artificial intelligence-based method –  the Gravitational Search Algorithm (GSA) based on Long-Short Term Memory (LSTM) network – and then carefully select optimal hyper-parameters, training algorithms, and activation functions to ensure accurate predictions.

The project, which will commence from July 2023 to June 2024, will deliver an advanced computational model for predicting the RUL and capacity of EV batteries. The performance of the LSTM-GSA algorithm will be evaluated using battery ageing data from the NASA dataset.

Upon the completion of this project, the researchers from Uniten and Universitas Pertamina aim to produce a new AI-based technique that can accurately predict the RUL and capacity of EV batteries. It will contribute to the reliability and safety of EVs by minimising the risk of premature battery failure.

The intensive and advanced research for EV-related technologies, coupled with the supporting policies and incentives by the governments in all countries, will certainly expedite the adoption of EVs in the next few years, such that it can contribute to a cleaner, greener, and more sustainable future.

The author is a senior lecturer at the College of Engineering and a member of the Institute of Sustainable Energy, Universiti Tenaga Nasional.

The views expressed here is the personal opinion of the writer and do not necessarily represent that of Twentytwo13.