Researchers at Qingdao University and collaborating institutions developed the framework as a specialized architecture for learning how battery performance changes over time. Their work, published in ENGINEERING Chemical Engineering, combines three technologies: coordinate feature attention, the Mamba2 state-space model, and a swiGLU-gated residual network. Together, these components are intended to identify the most informative signals in battery data, retain long-term degradation patterns, and convert complex learned representations into accurate predictions of the number of cycles a battery can continue operating.

The first component, the coordinate feature attention network, or CFAN, is designed to focus on where meaningful information occurs in a battery’s degradation history. Rather than treating all measurements and time points equally, the network analyzes information along two directions: the temporal dimension and the variable dimension. This allows it to generate attention weights that emphasize important degradation stages as well as critical health indicators. For example, the model can learn that a subtle change in a charging feature may become especially valuable when it occurs during a particular phase of the battery’s aging process. The bidirectional attention mechanism helps preserve both time-dependent and feature-dependent relationships.

At the center of CA-Mamba2 is the Mamba2 backbone, a selective state-space model developed to represent long sequences efficiently. Conventional recurrent networks can struggle to retain information over long operating histories, while Transformer-based systems often require substantial computational resources because of their attention calculations. Mamba2 takes a different approach by maintaining an internal state that evolves as new data arrive. Its parameters can be selectively adjusted according to the input, allowing the model to decide which information should be retained, updated, or suppressed. For battery prediction, this provides a way to connect early signs of degradation with conditions observed much later in the life cycle.

The framework’s final major element is the swiGLU-gated residual network, or SGRN. This component applies nonlinear gated transformations to the features extracted by the attention and state-space layers. Gating enables the network to regulate the flow of information, allowing useful signals to pass through while reducing the influence of less relevant patterns. Residual connections also preserve the original representations as the data move through the network, helping prevent the loss of information and improving training stability. The resulting decoder is intended to translate highly complex degradation features into an RUL estimate without discarding important characteristics learned earlier in the model.

The researchers tested CA-Mamba2 using public battery datasets from NASA, Tongji University, and Xi’an Jiaotong University. These datasets include different battery chemistries, operating conditions, and measurement configurations, providing a demanding test of whether the model can perform beyond a single experimental setting. On NASA’s B0005 test battery, the framework was evaluated using a single input variable and at prediction starting points corresponding to 50, 70, and 90 cycles. It produced mean absolute error values between 0.0089 and 0.0098, root mean square errors from 0.0140 to 0.0172, and coefficients of determination ranging from 0.949 to 0.989. The corresponding absolute prediction errors were approximately 1.2 to 2.3 cycles.

These results exceeded the performance of several widely used comparison models, including long short-term memory networks, Transformers, PathFormer, TimeMixer, and a baseline Mamba model. The advantages became even more pronounced on the Tongji University dataset, where the researchers used 17 variables, including voltage and current statistics and features extracted from the charging process. Across prediction starting points of 200, 300, and 400 cycles, CA-Mamba2 achieved mean absolute errors of 0.0014 to 0.0016 and root mean square errors of 0.0022 to 0.0024. Its R² values reached 0.9995 to 0.9998, while absolute errors remained between 2.4 and 2.5 cycles.

One particularly important finding was the model’s stability when the prediction starting point changed. In real-world battery management, an algorithm may be asked to estimate remaining life at very different moments, depending on how much operating data are available. Some competing approaches became substantially less accurate when the starting point shifted. The absolute error of TimeMixer, for example, increased from 12.0 cycles to 50.7 cycles across the tested starting points. CA-Mamba2, by contrast, maintained almost unchanged performance, suggesting that its combination of feature attention and long-range state modeling may help it remain reliable under changing observation conditions.

The model also demonstrated strong cross-dataset performance on the XJTU battery data. In tests designed to examine generalization across batteries and degradation stages, CA-Mamba2 produced mean absolute errors between 0.0049 and 0.0083, root mean square errors from 0.0066 to 0.0111, and R² values consistently above 0.986. Computational measurements further strengthened the case for practical use. On the Tongji dataset, CA-Mamba2 required 57.848 seconds for training and 0.476 seconds for inference, compared with 423.765 seconds for Transformer training and 1.528 seconds for inference. Ablation experiments showed that both CFAN and SGRN contributed to the final gains, with their combination producing the strongest overall results. The researchers say the framework could ultimately support battery management systems in electric vehicles and stationary energy storage, where accurate, fast, and adaptable lifetime predictions are essential for safety, efficiency, and lower maintenance costs.

Subject of Research: Not applicable

Article Title: A coordinate-aware Mamba2 framework for remaining useful life prediction of lithium-ion battery

News Publication Date: 13-Jul-2026

Web References: https://doi.org/10.1007/s11705-026-2687-x

References: DOI: 10.1007/s11705-026-2687-x

Image Credits: Higher Education Press

Keywords

Lithium-ion batteries, remaining useful life, battery degradation, artificial intelligence, Mamba2, state-space models, coordinate attention, battery management systems, electric vehicles, energy storage, machine learning

The implications of predicting the RUL of lithium-ion batteries are manifold. By accurately gauging how much life a battery has left, manufacturers can optimize battery management systems and enhance safety measures to prevent failures. Additionally, consumers of electric vehicles and portable devices stand to benefit through improved maintenance schedules and cost-effectiveness, ultimately leading to prolonged battery lifespan. The MWASFormer network employs a sophisticated blend of deep learning techniques that push the envelope of what is achievable in battery health diagnostics.

At the core of the MWASFormer model is an innovative architecture designed to harness a wealth of data collected on the performance metrics of lithium-ion batteries, including charge cycles, temperature fluctuations, and discharge rates. The model systematically analyzes these factors to create a predictive framework that not only estimates RUL but also provides invaluable insights into the underlying mechanisms of battery degradation. This feature sets MWASFormer apart from conventional methods, which often rely on simplistic models and fail to encapsulate the complexities involved in battery usage.

Moreover, the researchers have woven machine learning into the fabric of battery lifespan predictions, effectively merging data science with electrical engineering. The approach focuses on extracting nuanced patterns from historical battery performance data, thus enabling the model to learn from past experiences and enhance accuracy in predictions moving forward. With the unprecedented scale of data generated by battery operations, machine learning serves as a powerful ally in deciphering trends and predicting outcomes.

The methodology implemented by Zeng and colleagues involves a multi-faceted training process that incorporates both supervised and unsupervised learning techniques. By exploiting a diverse dataset representative of various operating conditions, the MWASFormer network is not just another theoretical model; it is a practical tool backed by empirical evidence. This statistical backbone lends credibility to the predictions, showcasing the model’s robustness under various scenarios.

In a world growing increasingly reliant on renewable energy resources, efficient battery management becomes crucial for the time-sensitive integration of solar and wind power into existing grids. The MWASFormer network emerges as a capable solution in this aspect, allowing stakeholders to manage energy storage more effectively. By forecasting battery longevity, energy providers can better align supply with demand, thereby optimizing grid operations and enhancing sustainability.

Another notable aspect of the research conducted is the way the MWASFormer model adapts to different battery chemistries and designs. Whether it’s lithium iron phosphate or lithium cobalt oxide, the network’s flexibility permits a tailored approach to battery management, making it applicable across a broad spectrum of technologies. This versatility broadens the scope of its implementation, positioning MWASFormer as a potential game-changer not just for consumer electronics but also for industrial applications.

Nonetheless, the model’s real-time applicability and integration into existing battery management systems will define its success. The research team emphasizes the importance of alignment between advanced predictive analytics and practical deployment conditions. To this end, they are exploring partnerships with battery manufacturers to facilitate the transition from laboratory findings to real-world applications.

Looking ahead, the researchers are committed to refining the model further, incorporating feedback from users and actual operational data. The iterative process of model enhancement means that the predictions will grow increasingly reliable, ultimately paving the way for smarter battery management systems globally. As energy storage needs evolve, MWASFormer stands poised to lead the charge in revolutionary battery lifespan forecasting.

The publication of their findings marks not only a milestone for the research team but also a significant leap forward for the broader energy storage community. As lithium-ion batteries continue to dominate the landscape, solutions like the MWASFormer network will become essential for understanding battery health and longevity more comprehensively. This advancement could even catalyze breakthroughs in energy technology that result in safer, longer-lasting batteries for future generations.

In summary, the MWASFormer network developed by Zeng, Jiang, and Wang represents a paradigm shift in the way we understand and predict the operational life of lithium-ion batteries. By integrating advanced machine learning techniques with rigorous data analysis, the model provides insights that could reshape battery management practices across various sectors. As industries emphasize sustainability through better energy use, approaches like the MWASFormer network can play an instrumental role in maximizing the efficacy of one of the most critical components of our energy infrastructure—lithium-ion batteries.

In conclusion, the journey toward optimizing battery life through predictive analytics represents a significant advance in materials science and battery engineering. The potential broader implications of accurate RUL predictions underscore the urgency for further research and development in the field. As the world shifts towards a future powered by sustainable energy solutions, innovations like those presented by Zeng and his team will undoubtedly help pave the way.

Subject of Research: Remaining useful life prediction of lithium-ion batteries

Article Title: Remaining useful life prediction of lithium-ion batteries based on MWASFormer network

Article References:
Zeng, L., Jiang, Z. & Wang, S. Remaining useful life prediction of lithium-ion batteries based on MWASFormer network. Ionics (2025). https://doi.org/10.1007/s11581-025-06794-0

Image Credits: AI Generated

DOI: 10.1007/s11581-025-06794-0

Keywords: lithium-ion batteries, MWASFormer network, remaining useful life, predictive modeling, battery management systems, machine learning, deep learning, energy storage, sustainability, battery longevity.