Predictive maintenance rests on a deceptively simple bargain. Algorithms trained on operational data learn to estimate a battery’s state of charge, state of health and remaining useful life, allowing operators to schedule interventions before failures occur rather than after. But the models behind these estimates are typically trained under a set of assumptions: that the battery will perform a defined duty, that environmental conditions will resemble those seen during training, and that the monitoring pipeline feeding data to the algorithm will stay consistent. The Perspective’s central claim is that all three assumptions are increasingly false for modern storage assets, and that the resulting mismatch, which the authors call operational drift, is a structural failure mode rather than an occasional nuisance.

The sources of drift are more varied than they might first appear. Application requirements evolve as grid markets change and new revenue streams emerge, so a system installed for peak shaving may later be dispatched for frequency regulation with very different charge and discharge profiles. Duty cycles shift accordingly, and research has shown that dynamic cycling can actually enhance battery lifetime, meaning that degradation trajectories can bend in either direction when usage patterns change. Environmental exposure varies too, from extreme cold climates that demand heating and cooling loads to be factored into system sizing, to temperature differences that strongly influence which degradation mechanisms dominate. Even the monitoring pipeline itself can drift, as sensor biases, random errors, low sampling rates and intermittent data unavailability alter the statistical character of the very signals the algorithms depend on.

To bring order to this complexity, the authors classify drift by its origin in evolving asset roles and then organize its effects into three layered functional mismatches. The first is a data-layer mismatch: the distribution of inputs seen in deployment no longer matches the training distribution, a phenomenon statisticians call covariate shift. The second is a model-layer mismatch, where the relationship the algorithm learned between features and outcomes no longer holds, for example when a new duty cycle activates degradation pathways such as lithium plating that were rare during training. The third is a temporal-layer mismatch, arising because degradation is not stationary in time; knee points on capacity-fade curves mark transitions from gradual to accelerated decline, and a model calibrated on early-life behavior can be badly wrong after such a transition.

Each mismatch layer, the authors argue, calls for a targeted transfer strategy rather than a one-size-fits-all fix. At the data layer, domain adaptation techniques can align source and target distributions, using statistical distances such as maximum mean discrepancy or Wasserstein distance to quantify and correct the shift. Adversarial variants go further, learning features that a domain classifier cannot distinguish, so that a model trained on laboratory cells or on one fleet can generalize to another. At the model layer, transfer learning approaches range from freezing a backbone network and attaching small trainable adapter layers, to physics-informed schemes that embed electrochemical knowledge, such as Arrhenius temperature dependence, directly into activation functions. Residual learning offers another elegant option: instead of learning the full mapping, the model learns only the difference between a physics-based reference prediction and observed behavior, which keeps adaptations small and interpretable.

The temporal layer demands perhaps the most delicate machinery, because adapting a model to new conditions can destroy what it already knows. The Perspective explicitly assesses candidate strategies against four failure risks: negative transfer, where adaptation to a new domain actually degrades performance; catastrophic forgetting, where learning new behavior erases old competence; poorly calibrated uncertainty, where the model’s confidence no longer reflects its true reliability; and low sample efficiency, where adaptation demands more target data than operators can realistically collect. Continual learning methods, including regularization-based approaches and bilevel-optimized schemes, have been shown to preserve earlier knowledge while absorbing new operating conditions, and meta-learning can adapt to a fresh battery with only a handful of target measurements.

What makes the argument timely is the sheer scale of the stakes. Grid-scale storage is expanding rapidly as renewable generation grows, and batteries increasingly provide multiple services simultaneously, from arbitrage and peak shaving to regulation and reliability support. Studies of vehicle-to-grid operation show that the dominant degradation mechanism determines whether extra energy throughput is worth the days of life lost, which means maintenance decisions are inseparable from market strategy. Meanwhile, retired electric vehicle batteries are flowing into second-life deployments where their histories are heterogeneous and often poorly documented, and where a model trained on first-life data may face a target domain it has never seen. The authors position transferable predictive maintenance as a life-cycle-spanning capability that connects manufacturing, first-life operation and second-life deployment into one continuous chain of health information.

The framework also carries implications that reach back to the factory floor. Research on transfer learning for battery cell manufacturing suggests that models trained during formation and early testing can predict lifetime outcomes, and that low-state-of-charge resistance measured immediately after manufacturing is a rapid predictor of how formation protocols will shape long-term durability. If such early-life signals could be carried forward through a battery’s passports and reused when the cell changes roles, the cold-start problem of second-life health estimation would shrink dramatically. Recent work on history-free prediction of second-life degradation trajectories from a single partial cycle, using interpretable mixture-of-experts models, hints at how much can be recovered even from sparse information, but the authors contend that systematic transferability would make such feats routine rather than exceptional.

None of this can be built on algorithms alone, and the Perspective is unusually concrete about the infrastructure required. The authors call for data infrastructures that preserve context across the life cycle, foundation models for batteries that encode broad electrochemical and operational knowledge and can be cheaply adapted to new tasks, and open benchmarks that test transferability under realistic drift rather than stationary laboratory conditions. Initiatives toward unified battery data descriptions and large-scale real-world datasets of retired cells point in this direction, as does work on federated learning, which allows fleets owned by different parties to collaborate on health estimation without sharing raw data. Privacy-preserving collaboration matters because the data that would make transferable maintenance most powerful, spanning manufacturers, fleet operators and recyclers, is exactly the data that commercial interests are most reluctant to pool.

The authors close with regulatory recommendations, arguing that standards should require calibrated uncertainty alongside point predictions and should encourage the audit dimensions their framework defines, so that operators can verify not just whether a model is accurate but whether it remains trustworthy as conditions drift. If they are right, the payoff extends well fewer failures and longer asset lives: batteries that can document their own health across changing roles would become easier to resell, repurpose and recycle, tightening the loop of a circular battery economy. In a decade when storage assets are expected to outlive the markets they were built for, the ability of their maintenance algorithms to move with them may prove as important as the chemistry inside the cells. Transferable predictive maintenance, the authors suggest, is the missing link between the batteries we build and the many lives we ask them to live.

Subject of Research: Transferable predictive maintenance for battery energy storage systems under operational drift

Article Title: Transferable predictive maintenance for battery energy storage systems

Article References: Tao, S., Liang, C., Tavallaey, S. S., Thulin, N., Wik, T., Zio, E., Pecht, M. G., Mo, H., & Zou, C. (2026). Transferable predictive maintenance for battery energy storage systems. Nature Reviews Electrical Engineering. https://doi.org/10.1038/s44287-026-00337-z

Image Credits: AI Generated

DOI: 10.1038/s44287-026-00337-z

Keywords: battery energy storage, predictive maintenance, transfer learning, operational drift, domain adaptation, continual learning, state of health, second-life batteries, uncertainty quantification, foundation models, grid storage, battery degradation

The research team assembled a benchmarking framework that put an unusually broad range of forecasting architectures through their paces under strictly identical experimental conditions. Rather than comparing models across different datasets, different preprocessing pipelines and different evaluation windows — a common weakness in the machine learning literature — the authors trained and tested every model on the same photovoltaic power time-series data, sourced from a publicly available Mendeley dataset of historical generation records. The contenders spanned three generations of forecasting technology: classical recurrent and convolutional deep learning models, modern transformer-based architectures, and the newest class of so-called foundation models, which are large networks pre-trained on vast collections of time series and then adapted to specific forecasting tasks.

In the classical corner sat three workhorses of sequence modeling. The long short-term memory network, or LSTM, uses gated memory cells to decide what information from past time steps should be carried forward, allowing it to capture slow-moving trends such as the daily solar cycle. The gated recurrent unit, or GRU, is a leaner relative that merges some of those gates for computational efficiency. Alongside them ran a one-dimensional convolutional neural network, or 1D CNN, which slides small filters across the input sequence to detect local patterns — sudden ramps in output, for instance — without the sequential processing that makes recurrent networks slow to train. These models have dominated solar forecasting papers for the better part of a decade, and they provided the baseline against which the newer architectures had to justify themselves.

The transformer family brought a fundamentally different mechanism to the table. Instead of reading a sequence step by step, transformers use multi-head attention, a mathematical operation that lets the model directly compare any two points in time and weigh their relevance to the prediction at hand. The study evaluated the Temporal Fusion Transformer, or TFT, an architecture designed for interpretable multi-horizon forecasting that combines attention with gating mechanisms and variable selection, and the iTransformer, a recent variant that reorganizes how the attention operation is applied across the dimensions of multivariate time-series data. The appeal of attention in solar forecasting is intuitive: a cloudy afternoon three days ago may matter more to tomorrow’s forecast than the smooth output of yesterday morning, and attention can learn to lock onto exactly those informative episodes.

The most forward-looking entrants were the foundation models. Time-MoE, introduced at the International Conference on Learning Representations in 2025, is a billion-scale time-series foundation model built on a mixture-of-experts design, in which different specialized sub-networks are activated for different inputs, allowing enormous model capacity without a proportional explosion in computation. Mamba4Cast takes a different route entirely, using state-space models — a mathematical framework for describing how a hidden internal state evolves over time — to achieve efficient zero-shot forecasting, meaning it can predict on data it has never seen without any task-specific training. These models represent the industry’s bet that forecasting, like language translation before it, will eventually be solved by enormous general-purpose networks rather than bespoke per-site models.

The headline result of the benchmark is a clear, graded improvement across architectural generations. The LSTM, the oldest model in the lineup, achieved a skill score of 0.29 and a root-mean-square error, or RMSE, of 0.158. In forecasting parlance, the skill score measures improvement over a naive persistence forecast — the assumption that the next value equals the current one — so 0.29 means the LSTM beat that naive baseline by a modest margin. The transformer-based and foundation-model families both pushed RMSE down to around 0.120 and 0.131 respectively, confirming that attention mechanisms and large-scale pre-training do translate into measurably better solar predictions. The gap may look small in absolute terms, but in grid operations, where forecasts drive billion-dollar scheduling decisions, even percentage-point improvements compound into significant economic and emissions savings.

The outright winner, however, was neither a pure transformer nor a pure foundation model, but a hybrid. The authors’ combined CNN-foundation-transformer architecture achieved a mean absolute error of 0.071, an RMSE of 0.106 and a normalized RMSE of 12.1, corresponding to a skill score of 0.64 against persistence forecasting — more than double the LSTM’s score. The logic of the combination is technically elegant. The convolutional front end extracts local, short-scale features such as abrupt irradiance ramps; the foundation component contributes temporal representations learned from an enormous diversity of time series, giving the model a robust sense of periodicity and seasonality; and the transformer’s attention layers then integrate these multi-scale signals, deciding which features from which time steps matter most for the forecast horizon. Visual analysis of the predictions confirmed that this architecture tracked day-to-day generation peaks and variability patterns with high fidelity, precisely the behavior that matters when operators must decide how much reserve capacity to hold.

Why should a hybrid beat its individual components? The answer likely lies in the complementary inductive biases each block contributes. Recurrent and convolutional layers impose strong structural assumptions about locality and continuity, which helps when data is limited; attention imposes few assumptions and can model long-range dependencies, but typically needs more data to shine; foundation models arrive pre-armed with generalized temporal knowledge that transfers across domains. Solar power data exhibits all of these regimes at once — smooth diurnal cycles, noisy cloud-driven fluctuations, and seasonal drift — so an architecture that layers multiple forms of temporal reasoning appears to capture the phenomenon more completely than any single mechanism. The finding echoes a broader trend in the field, where hybrid CNN-LSTM and CNN-LSTM-transformer designs have repeatedly outperformed monolithic models in prior studies of solar and load forecasting.

The practical implications reach well beyond academic leaderboards. Short-term PV forecasting is a linchpin of grid stability: system operators use these predictions to schedule conventional generation, manage battery storage, and participate in electricity markets. A forecast with a skill score of 0.64, as the best hybrid achieved, means substantially fewer surprise deficits on cloudy days and less curtailment of solar output on bright ones. Moreover, the study’s unified evaluation protocol is itself a contribution. Because every model faced identical data, preprocessing and metrics, the results offer a rare apples-to-apples comparison in a literature notorious for incomparable claims, giving practitioners a defensible basis for choosing architectures rather than relying on cherry-picked benchmarks.

The study also maps the road ahead. Foundation models like Time-MoE and Mamba4Cast performed strongly even though they were not designed specifically for solar data, suggesting that general-purpose temporal pre-training is a powerful starting point that domain-specific fine-tuning could push further. The authors’ work was supported by Al-Quds University’s Najjad Zeenni Faculty of Engineering, and the data and supplementary materials accompanying the paper provide a foundation for replication. As renewable penetration deepens, the contest between forecasting architectures will only intensify — and this benchmark suggests the future belongs not to any single paradigm, but to architectures that know how to make attention, convolution and pre-trained temporal knowledge work together.

Subject of Research: Machine learning architectures for short-term photovoltaic power time-series forecasting

Article Title: A systematic evaluation of classical, transformer-based, and foundation models for photovoltaic power time-series forecasting

Article References: Salman, D., Alzeer, I., & Isayed, W. (2026). A systematic evaluation of classical, transformer-based, and foundation models for photovoltaic power time-series forecasting. Neural Computing and Applications, 38(19), Article 764. https://doi.org/10.1007/s00521-026-12498-x

Image Credits: AI Generated

DOI: 10.1007/s00521-026-12498-x

Keywords: photovoltaic power forecasting, deep learning, transformer models, foundation models, LSTM, time-series forecasting, hybrid neural networks, Temporal Fusion Transformer, Time-MoE, Mamba4Cast, renewable energy, grid stability

The research team, led by Manisha Galphade and colleagues, started from a deceptively simple observation: the statistical character of wind speed data changes with the seasons. The mean, variance, distribution shape, autocorrelation, and extreme values of wind records all shift as atmospheric conditions move from one regime to another. A single general-purpose model, trained on a full year of data, tends to overfit the turbulent high-wind seasons and underfit the calmer ones. Classical statistical tools such as ARIMA and seasonal ARIMA assume linearity and stationarity, assumptions that wind power data routinely violates. Machine learning methods like random forests and support vector machines capture nonlinear relationships but struggle with long-term temporal dependencies. Even the deep learning tools that have transformed other fields face difficulties when the underlying signal changes its personality every few months.

Transformers, the architecture behind the current artificial intelligence boom, have been proposed as a solution. Models such as the Informer, the Temporal Fusion Transformer, and the Autoformer bring powerful attention mechanisms to time-series forecasting, and they have shown impressive results on long sequences. But the authors point out a practical catch: wind farms rarely produce the enormous, clean datasets that transformers crave. Sensor coverage is limited, measurements are noisy, and self-attention scales quadratically with sequence length, driving up training time and hardware demands. With many hyperparameters to tune and a tendency to overfit small, messy datasets, transformers can be an expensive and fragile choice for operational wind forecasting, particularly at installations that lack cloud-scale computing.

The team’s answer is a serial hybrid that chains two recurrent architectures together. A long short-term memory network, or LSTM, first processes the raw wind power sequence. LSTMs are built around memory cells governed by three gates—an input gate, a forget gate, and an output gate—that control what information is stored, discarded, and passed forward, allowing the network to learn long-range dependencies without suffering from vanishing gradients. The output of this LSTM is then fed into a bidirectional LSTM, or BiLSTM, which reads the refined representations in both forward and backward directions. Because the BiLSTM operates on already-filtered, higher-level features rather than noisy raw signals, the combination performs a kind of progressive feature abstraction: the LSTM extracts stable sequential patterns, and the BiLSTM layers contextual understanding on top of them.

Before any training begins, the raw data undergoes careful preprocessing. Missing values are filled using imputation methods, with a random forest regression imputer chosen for one dataset because it captures nonlinear relationships without assuming a particular data distribution. Outliers are detected by standardizing each data point and flagging values that deviate too far from the mean, and the cleaned features are then rescaled to the range between zero and one using min-max normalization. These steps matter more than they might appear: wind farm records are riddled with sensor dropouts, negative power readings caused by measurement artifacts, and spikes from unusual atmospheric events, and a forecasting model fed uncleaned data will happily learn the noise.

The researchers tested their approach on two real wind farms with very different characters. The first is the Yalova wind farm in western Turkey, a 54,000-kilowatt installation of 36 turbines whose supervisory control and data acquisition system recorded wind speed, direction, generated power, and theoretical power at ten-minute intervals throughout 2018, yielding more than 46,000 records. The second is the Sotavento wind farm in Galicia, Spain, with 24 onshore turbines and a capacity of 17,560 kilowatts, providing hourly meteorological and generation data for 2014. In both cases the data was split by season, with thirty-day windows drawn from winter, spring, summer, and autumn, and the models were evaluated using root mean squared error, mean absolute error, and the coefficient of determination.

The results reveal a striking seasonal fingerprint. At Yalova, spring and summer proved the easiest to forecast, with the hybrid model achieving coefficients of determination as high as 0.99 and its best summer performance at a lookback window of six time steps. Autumn, a transitional season mixing summer and winter behavior, produced moderate errors, while winter was hardest of all: volatility, sudden spikes and drops, and non-stationary behavior pushed the error metrics up and the explanatory power down to roughly 0.80. At Sotavento the same pattern emerged, with the optimal lookback window shifting from five steps in spring to three in summer and just one in autumn. The authors identify this systematic analysis of lookback window optimization as a central contribution, showing that no single window length serves all seasons and that adaptive, season-specific temporal context can substantially improve accuracy.

The hybrid model did not just beat its individual components. Against standalone LSTM and BiLSTM baselines, it reduced mean absolute error by 1.38 percent in spring, 12.1 percent in summer, about 4.67 percent in autumn, and 5.94 percent in winter. It also outperformed temporal convolutional networks, attention-based LSTMs, and transformer models across both datasets, and a Diebold-Mariano statistical test confirmed that most of these improvements were significant at the five percent level, with the gaps against TCN and transformer models highly significant. An ablation study reinforced the design choices: removing either component, reversing the layer order, changing the unit counts, dropping regularization, or altering the learning rate all degraded performance, sometimes dramatically, confirming that the specific architecture and its hyperparameters are genuinely well balanced rather than accidentally lucky.

Perhaps the most persuasive numbers concern efficiency. The proposed model contains just 10,913 parameters, occupies 171 kilobytes, and trained in about 43 seconds—faster than every competing model tested, including the much larger TCN with its 89,473 parameters. Its normalized accuracy of 96.34 percent topped the field, edging out the LSTM at 95.11 percent and the transformer at 92.03 percent. The authors attribute this to a favorable bias-variance trade-off: transformers and attention models carry representational capacity that moderate-sized wind datasets cannot exploit, so their extra parameters mostly buy overfitting risk, while the LSTM’s gating mechanism acts as an inherent noise filter that attention mechanisms sometimes lack. With linear computational complexity in sequence length rather than quadratic, the architecture is well suited to real-time forecasting systems and edge deployments where memory and compute are scarce.

The implications reach well beyond two Spanish and Turkish wind farms. Accurate seasonal forecasting underpins grid planning, reserve operation, and maintenance scheduling, and as wind penetration grows, the cost of forecast error grows with it. The study is candid about its limits: winter remains difficult for every model tested, and the authors suggest that incorporating external meteorological variables, attention mechanisms, and ensemble methods could push performance further. But the core message is a useful corrective to the prevailing enthusiasm for ever-bigger architectures. For seasonal wind power prediction on realistic, medium-sized datasets, a thoughtfully composed pair of recurrent networks—one reading time forward, one reading it both ways—can beat the giants while running on hardware that fits in a wind farm’s back pocket.

Subject of Research: Seasonal wind power forecasting using a hybrid LSTM-BiLSTM deep learning model tested on two wind farms

Article Title: Seasonal wind power forecasting using data-driven and computational intelligence techniques

Article References: Galphade, M., Dande, A., More, N., Nikam, V., & Hatkar, V. (2026). Seasonal wind power forecasting using data-driven and computational intelligence techniques. Results in Engineering, 32, Article 113122. https://doi.org/10.1016/j.rineng.2026.113122

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113122

Keywords: wind power forecasting, LSTM, BiLSTM, deep learning, seasonal variability, renewable energy, time series forecasting, transformer models, grid stability, Yalova wind farm, Sotavento wind farm, computational intelligence