Machine Learning-Driven Solutions for Battery Temperature Control in Electric Motorcycles

Researchers optimized a battery thermal management system (BTMS) using machine learning, combining air cooling and phase change materials (PCM) to efficiently manage lithium-ion battery temperatures. The system achieved high accuracy in temperature prediction and reduced costs, improving safety and performance in electric vehicles.

Researchers from the Jiangsu Engineering Research Center of New Energy Vehicle Energy Saving and Battery Safety at the Wuxi Institute of Technology in China have conducted a study on optimizing battery thermal management systems (BTMS) for electric motorcycles using machine learning. With the increasing adoption of electric vehicles as a solution to mitigate environmental pollution, lithium-ion batteries have become central to extending the driving range of these vehicles. However, lithium-ion batteries are highly sensitive to temperature variations, which can lead to reduced capacity or, in extreme cases, dangerous thermal runaway situations. Managing battery temperature effectively is critical to ensuring both the safety and performance of electric motorcycles.

Balancing Battery Safety and Efficiency with Advanced Cooling

The study focuses on a battery thermal management system that integrates air cooling and phase change material (PCM) cooling to maintain battery temperatures within an optimal range. The research emphasizes that an effective BTMS should not only control temperature but also minimize energy consumption and manufacturing costs. The researchers explore how different parameters such as PCM thickness, PCM height, and air velocity affect battery temperature, and they use machine learning algorithms to optimize the BTMS structure for cost and performance. Their findings indicate that while increasing PCM thickness and air velocity improves heat dissipation, the benefits diminish after a certain point.

The Challenge of Lithium-Ion Battery Temperature Sensitivity

The study begins by analyzing the limitations of lithium-ion batteries compared to lead-acid batteries, which were once more commonly used in electric motorcycles. Lithium-ion batteries offer superior energy density and charging efficiency, making them the preferred choice today. However, their sensitivity to temperature poses a challenge, as both high and low temperatures can impact their performance. The optimal operating temperature for lithium-ion batteries is between 20°C and 40°C, with temperature differences of more than 5°C to be avoided. When the temperature exceeds these limits, thermal runaway can occur, leading to potential safety risks for the vehicle and its driver.

Exploring Cooling Techniques: Active and Passive Methods

Active and passive cooling methods are commonly used in BTMS to address this issue. Active cooling includes air or liquid cooling, while passive cooling involves the use of PCMs or heat pipes. The researchers highlight recent studies that have successfully combined these two cooling methods to create more efficient thermal management systems. The integration of PCM into air cooling systems is particularly effective in absorbing heat and maintaining stable temperatures. However, the challenge lies in optimizing the amount of PCM and air cooling required to achieve the best results without increasing costs or energy consumption unnecessarily.

Optimizing BTMS Structure with the Black Kite Algorithm

To tackle this challenge, the researchers employed the Black Kite Algorithm (BKA), a bio-inspired optimization algorithm, to refine the structure of the BTMS. The BKA was used to minimize the cost of the system while ensuring that the battery temperature remained below a critical threshold. The study found that increasing the thickness of the PCM layer improves the system's ability to dissipate heat, but with diminishing returns after a certain point. Similarly, increasing the height of the PCM enhances cooling efficiency at low air velocities but can hinder cooling at higher air velocities due to reduced direct heat exchange between the battery and the air.

Through experimentation and simulation, the optimal PCM height was determined to be 35 mm, resulting in a BTMS cost of 0.073 USD per battery. This cost includes the material expenses for the PCM as well as the energy consumption of the cooling fans. In addition to structural optimization, the study also developed a battery temperature prediction model using machine learning. A combination of a Convolutional Neural Network (CNN) and a Self-Attention (SA) mechanism was employed to predict battery temperature during operation. This model was further optimized using the BKA to improve accuracy.

Achieving Precision with the BKA-CNN-SA Prediction Model

The BKA-CNN-SA model was trained on experimental data and validated using a test set. The results showed that the model could predict battery temperature changes with a maximum error of only 0.45°C. This high level of accuracy was achieved thanks to the model's ability to learn complex relationships between various factors such as current, voltage, air velocity, and ambient temperature. The researchers also compared their model's performance to other machine learning models and found that the BKA-CNN-SA model outperformed others in terms of accuracy, with an R2 score of 0.998, a mean absolute error of 0.104, and a root mean square error of 0.131.

The study concludes that combining PCM and air cooling in BTMS offers a cost-effective and efficient solution for managing battery temperatures in electric motorcycles. By optimizing the system's structure and employing advanced temperature prediction models, the researchers provide valuable insights for designing BTMS that balance cost, performance, and safety. Their findings offer a practical approach to improving battery management systems for electric vehicles, contributing to the ongoing efforts to enhance the performance and safety of environmentally friendly transportation.