If mobile phone battery manufacturers can pre-judge which batteries can be used for at least two consecutive years, then they can sell these batteries to mobile phone manufacturers and sell the rest to less demanding equipment manufacturers. New research shows that battery manufacturers can do this. This technology can be used not only to classify the manufactured batteries, but also to help new battery designs enter the market faster. Scientists at Stanford University, the Massachusetts Institute of Technology and the Toyota Research Institute found that comprehensive experimental data and artificial intelligence revealed the key to accurately predicting the life of lithium-ion batteries before they begin to decline. After the researchers trained their machine learning model with hundreds of millions of battery charge and discharge data points, the algorithm predicted how many cycles each battery could last based on voltage drops and some other factors in the early cycle. The algorithm can distinguish the long life and short life of the battery based only on the first five charge and discharge cycles. Here, 95% of the predictions are correct. This machine learning method was published in the March 25 issue of "Nature Energy" (Nature Energy) magazine, it can accelerate the design and development of new batteries, reduce production time and cost, and other applications. Researchers have published the data set-the largest of its kind. The standard method for testing new battery designs is to charge and discharge the battery until the battery fails. Peter Attia, a PhD student in materials science and engineering at Stanford University, said that due to the long battery life, this process may take months or even years. This is an expensive bottleneck in battery research. This work was carried out at the Center for Data-Driven Design of Batteries, which is an academic and industrial cooperative institution integrating theory, experiment and data science. Researchers led by William Chueh, assistant professor of materials science and engineering at Stanford University, conducted battery experiments. The MITs team, led by Richard Braatz, a professor of chemical engineering, completed machine learning. Kristen Severson is the co-lead author of the study, and he completed his Ph.D. in chemical engineering at the Massachusetts Institute of Technology (MIT) last spring. Optimize fast charging One focus of the project is to find a better way to fully charge the battery in 10 minutes. This feature may accelerate the large-scale application of electric vehicles. To generate the training data set, the research team charged and discharged the batteries until the end of each battery's life, and they defined it as a 20% capacity loss. In the process of optimizing fast charging, the researchers wanted to find out whether the answer to the battery question can be found in the early cycle information? Advances in computing power and data generation have recently enabled machine learning to accelerate the completion of various tasks. Bratz said that this includes predictions of material properties. Our results show that we can predict the behavior of complex systems in the distant future. Generally speaking, the capacity of lithium ion batteries is stable for a period of time. Then it will drop sharply. As most consumers in the 21st century know, there is a big difference in the drop point. In this project, the battery can be used 150 to 2300 times. This change is partly due to the testing of different fast charging methods, but it is also determined by the manufacturing differences between the batteries. Patrick Herring, one of the authors of the research report and a scientist at the Toyota Research Institute, said: "In this work, we reduced one of the most time-consuming steps-battery testing-by an order of magnitude . " Possible uses Attia said the new method has many potential applications. For example, it can shorten the time to verify new batteries, which is especially important considering the rapid development of materials. With this rapid sorting technology, batteries with shorter life spans—too short for cars—can be used to power street lights or as backup batteries in data centers. Recyclers can find suitable batteries from the used battery packs of electric vehicles, and the capacity of these batteries is enough to be used for ladders. Another possibility is to optimize battery manufacturing. Attia said the final step in manufacturing batteries is molding, which may take days to weeks. Using our method can greatly shorten this process and reduce production costs. The researchers are now using their model to optimize the high-speed charging method within 10 minutes. They say this will reduce the battery charging time by more than 10 times. (Originally from: Daily Science China New Energy Network Synthesis) Emergency Lighting System,Emergency Lights For Home Power Failure,Rechargeable Emergency Light,Led Emergency Lights Foshan Nai An Lighting Electric Co.,ltd , https://www.fsnaipsled.com