Can AI unlock electric vehicle battery reliability?

The recent Q4 2023 – Automotive industry demand forecast report from the Advanced Propulsion Centre UK identifies a number of themes for development in the field of battery electric vehicles (BEVs) that lead to lower cost and improved performance and reliability. 

Structural developments will lead to increased gravimetric and volumetric densities and other aspects have the potential to increase structural integrity. One particular challenge is to be able to monitor and manage the long term health and performance of batteries. 

The proposal is to use AI and wireless cloud connectivity to predict and maintain the health of battery cells and protect from catastrophic failure and degradation. 

Wireless connectivity will allow the collection of battery health data from a large number of vehicles, and analysis of this data by machine learning and prediction of anomalies using AI will help maintain battery health. On-board battery management systems can apply this learning to individual vehicles. 

The improvement of reliability will have an effect on the long term values of the vehicles by giving better insight into the state of battery health. It also provide information for insurers to evaluate the potential risk and adjust premiums accordingly. Overall, these developments using the capabilities of AI and machine learning could have a significant impact on one of the current problems affecting BEV uptake: that of uncertain product lifetime and long term values. 

In the recent Marks & Clerk AI Report 2023, we noted that the trend for AI patent filings is increasing. Transportation is one of the top five industry sectors for AI patent filings. It currently has the highest allowance rate, at over 60%. Therefore, not only does AI and machine learning form a key part of the technology development pathway for BEVs, it is also an area in which patent protection is a viable route for protection of the investment in technology by the relevant parties. 

Where AI is used to provide a direct input to a control process, it is usually straightforward to define a technical contribution necessary for patent protection. Current EPO Guidelines for Examination (Section G-II, 3.3.2) also recognise that the data of a simulation or model may have a “potential technical effect” that will be produced when the data are used according to an intended technical use.  By careful patent drafting, it should therefore be possible to secure useful patent protection for at least some of these BEV developments.