The physics of the problem is deceptively simple. In the torque-speed plane, motoring occupies the quadrants where torque and rotation share a sign, so mechanical power, given by P = Tω, is positive. Regenerative braking flips that product negative: the motor opposes rotation and acts as a generator, and the inverter routes the resulting electrical power back into the battery if charging is permitted. Whatever braking demand the generator cannot absorb must be supplied by friction brakes, which respond on hydraulic timescales that are often slower than the near-instantaneous collapse of motor torque. When a battery limit falls during a stop, the friction system may not yet have built pressure, and for a brief window the car simply has less braking torque than the driver requested.

The review’s central insight is that battery charge acceptance is not a single number a controller can look up. State of charge describes stored charge relative to a reference capacity, but it does not by itself define allowable charging power. A battery management system must weigh voltage, temperature, and charging duration as well. Temperature is particularly treacherous because ambient and cell temperatures are different control inputs: cell temperature depends on recent operation and thermal management, and a cold battery may accept more charging power as it warms while further heating can reduce that limit. Studies cited in the review, including work by Steinstraeter and colleagues on low-temperature range effects and Zhao and colleagues on energy flows across ambient temperatures, show how strongly these conditions shape recovery, without defining a universal temperature threshold for all batteries.

Just as important is the review’s separation of immediate battery protection from long-term aging. Voltage, current, and temperature limits restrict charging power at a given instant, while aging models estimate damage accumulated over an operating history. Wu and colleagues combined charging-power limits with a semi-empirical aging model in a hierarchical controller, but the authors stress that reducing a modeled aging cost can never justify exceeding a hard charging limit, and that operation within the limit can still produce different aging outcomes over time. At the materials level, degradation is associated with coupled cathode failure mechanisms including structural instability, interfacial side reactions, and particle cracking, which provides context for aging assessment even though the specific contributions under regenerative charging conditions still require validation.

Perhaps the most counterintuitive finding the review highlights is that regenerative charging pulses do not necessarily shorten battery life. In cell experiments cited in the review, commercial lithium-ion cells tested under dynamic loads and varying regeneration levels showed that the benefit of shallower discharge outweighed the damage associated with short charging pulses. The authors are careful to note this result is specific to the tested cells and conditions, and that new battery designs need fresh pulse-charge and aging data before such conclusions can guide control. Lower charging current alone, they argue, does not establish a battery-life benefit; aging comparisons must consider the full charge-discharge history rather than peak regenerative current in isolation.

On the mechanical side, the review is equally blunt about what regeneration cannot do. Regenerative and friction braking act through the same tire contact, so swapping one for the other never increases road grip. Near the grip limit, wheel-slip control restricts braking torque regardless of where it comes from, and on split-friction surfaces the left and right tires face different limits while cornering steals force from braking. The authors draw a sharp distinction between hard limits and control targets: a desired slip value is a target, an allowed slip range is a constraint, motor efficiency is an optimization goal, and maximum charging power is a limit. If tire grip is already fully used, adding friction torque cannot produce the requested deceleration; total torque must instead be limited to control slip.

Against this backdrop, the review compares the major families of brake control. Rule-based allocation caps regenerative torque and hands the remainder to friction brakes, offering transparency but suffering from a static torque split that says nothing about the transfer between actuators. Fuzzy control can encode nonlinear allocation rules without a full system model, but its rule base still needs a separate check of hard limits because untested combinations of battery temperature, voltage, speed, and grip may fall outside the tuning conditions. Model predictive control predicts system response over a short horizon and selects commands subject to constraints, with variants ranging from hybrid-bus applications to explicit MPC tested with hydraulic hardware in the loop, and studies that included actuator dynamics validated on a research vehicle. Learning-based methods, from reinforcement-learning deceleration planning to neural-network brake control and combinations of fuzzy logic with deep reinforcement learning, promise smoother plans but still require command limiting and fallback actions verified in closed-loop tests.

The review also examines how production vehicles have approached the driver-facing side of the problem. Dual-motor layouts such as the Porsche Taycan Turbo and Turbo S, which document front and rear motors with regenerative and hydraulic brake blending, gain more torque-allocation choices, though two motors do not double the charging power the battery can accept. One-pedal control, exemplified by the Tesla Model Y manual’s description of regeneration after accelerator release and automatic friction-brake compensation when regeneration is limited by a cold or full battery, reduces pedal switching in routine driving but demands careful calibration of pedal-release response, low-speed friction blending, and predictable behavior when regeneration changes. Porsche’s documented approach of coasting on accelerator release with recovery mainly through brake-pedal operation offers a different philosophy, and the authors note that relative energy use and braking feel between these interfaces require matched tests rather than assumptions.

At the heart of the paper is a proposal to measure what happens when regenerative capability changes mid-brake. If a reduction in charging capability can be predicted, friction pressure can build while regenerative torque is reduced, keeping their sum close to the request. A sudden battery limit, by contrast, may force an immediate torque reduction before friction torque is ready. The authors define torque-deficit measures for battery-limited transfers with sufficient tire grip: the peak deficit in newton-meters, its duration in seconds, and its time integral in newton-meter-seconds, all computed as the positive difference between requested and delivered wheel-side braking torque, with the analysis window and noise threshold stated. During ABS intervention, they recommend using the slip-limited torque reference rather than the raw driver request, so that a necessary safety action is not miscounted as a tracking failure.

Validation, the review concludes, must progress from model checks through hardware-in-the-loop tests to controlled vehicle trials, with each study identifying which battery, brake, tire, and driver components are real and which are simulated. Proposed test cases combine battery conditions with braking events: sustained deceleration under restricted charging, sudden regeneration loss, low-speed stopping, and ABS intervention, with split-friction and cornering tests added for any claims about yaw or lateral stability. Energy-recovery figures are only comparable when measurement points and denominators match, since wheel, inverter, and battery-terminal energy include different losses, and a higher final state of charge says nothing about state of health. Looking forward, the authors argue that battery management should hand brake control not just a power limit but its timestamp and an uncertainty margin, and that new fast-charging packs such as CATL Shenxing Pro, BYD Blade Battery 2.0, and Geely Short Blade cells warrant fresh charge-acceptance and pulse-aging data before their limits are trusted in braking control. Better charge acceptance can raise recovery when the battery is the active constraint, but motor capability, tire grip, and braking demand will always keep the final word.

Subject of Research: Coordination of regenerative braking control with battery charge acceptance, motor limits, and tire grip constraints in electric vehicles

Article Title: Regenerative braking control for electric vehicles with battery and safety constraints

Article References: Changyong, S., Wenfeng, Z., & Zhu, M. (2026). Regenerative braking control for electric vehicles with battery and safety constraints. Frontiers in Mechanical Engineering, 12, Article 2005334. https://doi.org/10.3389/fmech.2026.2005334

Image Credits: AI Generated

DOI: 10.3389/fmech.2026.2005334

Keywords: regenerative braking, electric vehicles, battery charge acceptance, model predictive control, wheel-slip control, battery aging, brake blending, torque allocation, one-pedal driving, dual-motor vehicles, friction brakes, energy recovery

The choice of cobalt disulfide, or CoS2, is no accident. Transition metal sulfides have become darlings of the pseudocapacitance world because their sulfur atoms help shuttle electrons quickly through the crystal lattice while the metal centers host fast, reversible surface redox reactions. CoS2 in particular offers metallic-like conductivity compared with many oxide alternatives, which means charge can move in and out of the active material without the crippling internal resistance that plagues poorer conductors. The catch is that sulfide electrodes often degrade over thousands of cycles, so the team paired CoS2 with neodymium oxide, a rare earth sesquioxide known for chemical stability and a knack for improving charge-transfer kinetics when incorporated into composite electrodes.

The synthesis was tuned with care. The researchers settled on a confirmed cobalt-to-neodymium molar feed ratio of 2:1, a proportion that produced a composite dominated by mesopores, pores in the two-to-fifty-nanometer range that are large enough for electrolyte ions to penetrate rapidly yet numerous enough to expose enormous internal surface area. Gas adsorption measurements using the Brunauer-Emmett-Teller method put that surface area at 213.21 square meters per gram, a figure comparable to many dedicated activated carbons but achieved here in a redox-active composite. That combination of open pore architecture and electrochemically active components is what allows an electrode to deliver charge both quickly and in quantity.

When tested in a three-electrode configuration, the CoS2/Nd2O3 composite delivered a specific capacitance of 1961 farads per gram at a modest current density of 1 ampere per gram. More impressive still is what happened under stress: at a punishing 20 amperes per gram, twenty times the current, the electrode still managed 1401 farads per gram. That rate capability, retaining roughly 71 percent of its low-rate capacitance, is the signature of an electrode where ions do not get bottlenecked in tortuous pores and electrons do not get stranded in resistive regions. After 10,000 charge-discharge cycles, the material retained 87.6 percent of its capacitance, evidence that the neodymium oxide component and the composite structure helped stabilize the sulfide against the mechanical and chemical wear that typically erodes such electrodes.

On the other side of the cell, the team turned to carbon nanotubes, the cylindrical graphitic filaments prized for their continuous one-dimensional conductivity. Pristine nanotubes store charge mostly through the electric double-layer mechanism, physically separating ions at their surface, which is fast but modest in capacity. The researchers treated the nanotubes with sulfur, creating sulfur-related surface species while preserving the continuous conductive network. The result was an S-CNT negative electrode delivering 431 farads per gram at 1 ampere per gram, 28.7 percent higher than untreated nanotubes measured under identical conditions. Even at 20 amperes per gram the S-CNTs held onto 81.0 percent of their capacitance, confirming that the sulfur modification added pseudocapacitive character without sacrificing the speed that makes carbon electrodes attractive.

Putting the two together yielded an aqueous asymmetric supercapacitor, a device architecture that exploits the different operating voltage windows of the two electrodes to push the total cell voltage well beyond what either electrode could sustain alone. Water-based electrolytes are inherently safer and cheaper than the organic solvents used in many commercial supercapacitors, though they traditionally cap the voltage window near 1.6 volts before water begins to split. The asymmetric design squeezes maximum benefit from that constraint by balancing a high-capacity battery-like positive electrode against a fast, stable carbon negative electrode.

The full two-electrode device held 73.3 percent of its capacitance when the current density was raised from 1 to 15 amperes per gram, and it survived 10,000 cycles with approximately 95.0 percent retention, a durability figure that stands out even among laboratory-scale aqueous devices. The authors are candid about the engineering subtleties involved. A retrospective first-order charge-balance analysis, which examines whether the charges stored on the positive and negative electrodes are properly matched, indicated that further optimization of the electrode mass ratio would be beneficial. In asymmetric supercapacitors, mismatched electrode masses leave one electrode underutilized, wasting capacity and shortening cycle life, so this kind of self-critical analysis points directly at the next round of improvement.

The motivation behind the work is refreshingly concrete. Short-duration and repetitive power fluctuations from train acceleration, braking, and regenerative-energy recovery demand storage devices with rapid response, high-rate capability, and stable cycling, and the authors frame their material pair as a laboratory-scale platform for exactly that kind of short-duration power buffering in rail transit. Regenerative braking energy is notoriously difficult to capture because it arrives in sharp spikes rather than smooth flows; batteries absorb it slowly and degrade quickly, while conventional double-layer capacitors lack the energy density to make a dent in a metro line’s power budget. A pseudocapacitive composite that can swallow thousands of farads per gram and release them at 20 amperes per gram, cycle after cycle, fits the profile of what wayside and on-board storage systems need.

The team is equally clear about what remains to be done before such a device could ride the rails. They note that composition optimization, complete-device impedance characterization, and application-specific pulse validation would further strengthen the assessment for rail-transit use. Impedance characterization, in particular, matters because real transit duty cycles involve not just repeated pulses but temperature swings, vibration, and years of continuous service, all of which show up first as changes in a device’s internal resistance. The authors also report that no datasets were generated or analyzed beyond the study itself, and the work was supported by the Key Project of the 2025 Educational Science Program of Heilongjiang Province under the 14th Five-Year Plan.

Still, the numbers tell a compelling story. A positive electrode approaching 2000 farads per gram, a sulfur-doped carbon negative electrode beating its untreated counterpart by nearly 29 percent, a full cell that keeps 95 percent of its capacitance over ten thousand cycles, and an honest roadmap for refinement together sketch a credible path from bench chemistry to power electronics. If subsequent work tightens the mass balance and validates the device under realistic pulse profiles, the humble pairing of a cobalt sulfide, a rare earth oxide, and a dash of sulfur on carbon nanotubes could find itself quietly smoothing the electricity of the world’s metro systems, one braking train at a time.

Subject of Research: CoS2/Nd2O3 composite and sulfur-doped carbon nanotube electrodes for asymmetric supercapacitors aimed at rail-transit power buffering

Article Title: Preparation of CoS2/Nd2O3 composites and S-CNTs and their electrochemical performance in asymmetric supercapacitors

Article References: Zhang, W., Lan, Y., Zheng, X., Zhang, Y., & Shen, W. (2026). Preparation of CoS2/Nd2O3 composites and S-CNTs and their electrochemical performance in asymmetric supercapacitors. Ionics. https://doi.org/10.1007/s11581-026-07486-z

Image Credits: AI Generated

DOI: 10.1007/s11581-026-07486-z

Keywords: asymmetric supercapacitor, CoS2/Nd2O3 composite, sulfur-doped carbon nanotubes, pseudocapacitance, electrode materials, regenerative braking, rail transit energy storage, mesoporous materials, aqueous electrolyte, cycle stability, rate capability, electrochemistry

The team’s starting point was a rich dataset of 72 real-driving trips performed with a BMW i3, a compact electric vehicle whose battery and heating circuit data were recorded under genuine operational conditions. This kind of data is gold for battery researchers. Synthetic drive cycles, such as the standardized profiles used in certification tests, smooth away the stop-and-go chaos, temperature swings, and driver behavior that dominate everyday motoring. Previous comparative studies have shown that battery degradation patterns on synthetic cycles can differ markedly from those observed in the wild, which means control strategies tuned in the laboratory often underperform on the street. By anchoring their models to 72 authentic journeys, Chen and Zhou ensured that every prediction and optimization they made reflected the messy reality of daily driving.

The first technical step was validation. The researchers constructed detailed vehicle and heating circuit models and checked them against the recorded trip data, confirming that their simulations reproduced the thermal and electrical behavior of the actual car. With trustworthy models in hand, they then developed strategies for a clever energy trick: using regenerative braking energy to heat the cabin. In an electric vehicle, braking normally wastes kinetic energy as heat at the friction pads, but regenerative systems recover part of it as electricity. Routing that recovered energy into cabin heating rather than drawing it from the traction battery eases the load on the cells, and the study reports that such strategies improved both the vehicle’s driving range and battery life. It is a double win: passengers stay warm, and the battery endures fewer stressful charge-discharge excursions.

From there, the study pivoted to machine learning. The researchers framed a prediction problem in which environmental variables, vehicle variables, battery variables, and thermal system variables serve as inputs, while the target outputs are the battery’s final State of Charge and the distance traveled in kilometers. In plain terms, they wanted an algorithm that could look at the conditions of a trip and accurately forecast how much energy would remain in the battery and how far the car would go. Four candidate models were systematically compared: XGBoost, AdaBoost, CatBoost, and K-Nearest Neighbor. Each was assessed with a rigorous protocol, an 80/20 train-test split combined with five-fold cross-validation, which guards against the common pitfall of an algorithm that merely memorizes its training data.

Among the four contenders, CatBoost, a gradient-boosting method developed originally to handle categorical features gracefully, emerged as the strongest baseline performer. On the held-out test set it achieved a coefficient of determination, or R², of 0.7260, a root mean square error of 0.0734, and a mean absolute error of 0.0592, while also posting the highest cross-validation accuracy of the group with an R² of 0.802. For readers unfamiliar with these metrics, R² measures the fraction of variance in the outcomes that the model explains, with 1.0 being perfect, while the error metrics quantify the typical size of the model’s mistakes. A baseline R² above 0.72 on real-world driving data is respectable, but the authors were not satisfied.

The study’s most distinctive contribution lies in what came next: metaheuristic optimization. Metaheuristics are search algorithms inspired by natural processes, from the flocking of birds to the foraging of insects, that explore vast parameter spaces far more efficiently than brute-force grid searches. Chen and Zhou deployed three such algorithms to fine-tune the CatBoost model’s hyperparameters, the internal settings that govern how aggressively the model learns and how it balances bias against variance. The winner was the Horned Lizard Optimization algorithm, a relatively new nature-inspired method, which when coupled with CatBoost produced the study’s best-performing framework, dubbed HLO-CatBoost.

The improvements were substantial. The optimized HLO-CatBoost framework achieved a test R² of 0.8408, with the root mean square error falling to 0.0559 and the mean absolute error to 0.0434. In other words, hyperparameter tuning by the horned lizard algorithm lifted the explained variance by more than eleven percentage points while cutting the typical prediction error by roughly a quarter compared with the untuned baseline. That gap matters in practice: a battery management system that predicts remaining charge and range more accurately can make smarter decisions about when to heat the cabin, when to limit power, and how to keep cells inside their optimal temperature window, directly translating into extra kilometers per charge and slower long-term degradation.

Crucially, the authors did not stop at headline accuracy numbers. They subjected their framework to a battery of further analyses designed to probe its robustness and trustworthiness. Environmental scenario testing examined how the model behaves across different ambient conditions, a critical check for a system intended to operate from freezing winters to hot summers. Uncertainty analysis quantified the confidence one can place in individual predictions. Computational efficiency assessments confirmed that the framework is fast enough for practical deployment rather than being a laboratory curiosity. A Williams plot, a standard diagnostic from chemometric modeling, was used to identify outliers and define the model’s applicability domain, flagging any predictions that fall outside the territory the model learned from. Finally, feature importance analysis opened the black box, revealing which input variables most strongly drive the predictions of final state of charge and travel distance.

This emphasis on interpretability and validation distinguishes the work from much of the machine learning literature applied to batteries, where a single impressive accuracy figure is often the whole story. Battery thermal management systems, or BTMSs, are safety-critical components: they must prevent thermal runaway, a catastrophic failure mode in which a cell overheats and triggers a chain reaction, while simultaneously squeezing out every possible increment of efficiency. A model that cannot explain itself, or that fails unpredictably at the edges of its training distribution, is a liability in that context. By combining rigorous cross-validation, outlier diagnostics, scenario analysis, and feature attribution, Chen and Zhou have assembled the kind of evidence package that engineers and regulators need before trusting an intelligent management system with a live vehicle.

The broader implications reach across the electric vehicle industry. Range anxiety remains one of the most cited barriers to EV adoption, and a meaningful share of that anxiety stems from the unpredictability of range in extreme weather, when heating and cooling loads can slash advertised figures. Machine learning frameworks trained on genuine driving data, and sharpened by metaheuristic optimization, offer a path toward management systems that anticipate those losses and compensate for them, whether by pre-heating the battery while the car is still plugged in, by harvesting braking energy for the cabin, or by modulating thermal loads in real time. The study’s authors, who collected and analyzed the data and evaluated and edited the manuscript, report no competing interests, and their work was supported by research programs at Chongqing institutions along with technical support from the marketing department of Chongqing Changan New Energy Automobile. As electric fleets grow and software-defined vehicles become the norm, the marriage of real-world data, gradient boosting, and nature-inspired optimization may well become a standard ingredient in the batteries that move us, turning every braking event and every degree of ambient temperature into information that helps a car go farther on the same charge.

Subject of Research: Machine learning optimization of electric vehicle battery thermal management using real-world driving data

Article Title: Optimizing automotive battery thermal regulation strategies for enhanced efficiency in real driving cycles

Article References: Chen, B., & Zhou, C. (2026). Optimizing automotive battery thermal regulation strategies for enhanced efficiency in real driving cycles. Ionics. https://doi.org/10.1007/s11581-026-07474-3

Image Credits: AI Generated

DOI: 10.1007/s11581-026-07474-3

Keywords: electric vehicles, battery thermal management, real driving cycles, CatBoost, Horned Lizard Optimization, machine learning, state of charge, regenerative braking, BMW i3, hyperparameter tuning, range prediction, battery degradation