Hui Shi

Capacity degradation in lithium-ion batteries may exhibit nonlinear decay, capacity regeneration, and local stochastic fluctuations under the combined effects of temperature variations, changes in discharge rate, and nonstationary measurement noise. Existing models usually treat temperature and discharge rate independently or combine their effects linearly, which cannot adequately capture rate-induced heat accumulation and its feedback on degradation. Conventional adaptive filters may also misattribute short-term measurement disturbances to process noise. This study therefore proposes an online remaining useful life (RUL) prediction framework that integrates a temperature-discharge-rate-coupled Wiener degradation model with a time-varying adaptive unscented Kalman filter (TV-AUKF) based on hierarchical noise decoupling. The discharge rate is embedded in the Arrhenius exponential term and combined with Peukert's relationship to construct a nonseparable coupled drift rate that describes accelerated degradation under temperature-rate interactions. Measurement-noise adaptation is separated from model-parameter updating: a relative-volatility-derived time-varying factor adjusts the measurement-noise covariance and detects abrupt operating-condition changes, thereby suppressing the effects of capacity regeneration and local anomalies on state estimation. Model parameters and the Wiener diffusion intensity are updated by a sliding-window Bayesian estimator using maximum likelihood estimation (MLE) priors, time-decay weights, and filtering-uncertainty weights; the resulting process-noise covariance is fed back to the UKF time prediction at the next step. Validation on the complementary CALCE, NASA PCoE, and Wenzhou Randomized datasets shows that the proposed method improves degradation-modeling accuracy under controlled stress, the robustness of online estimation under nonstationary noise, and the generalizability of RUL predictions under dynamic operating conditions.

Capacity degradation in lithium-ion batteries may exhibit nonlinear decay, capacity regeneration, and local stochastic fluctuations under the combined effects of temperature variations, changes in discharge rate, and nonstationary measurement noise. Existing models usually treat temperature and discharge rate independently or combine their effects linearly, which cannot adequately capture rate-induced heat accumulation and its feedback on degradation. Conventional adaptive filters may also misattribute short-term measurement disturbances to process noise. This study therefore proposes an online remaining useful life (RUL) prediction framework that integrates a temperature-discharge-rate-coupled Wiener degradation model with a time-varying adaptive unscented Kalman filter (TV-AUKF) based on hierarchical noise decoupling. The discharge rate is embedded in the Arrhenius exponential term and combined with Peukert's relationship to construct a nonseparable coupled drift rate that describes accelerated degradation under temperature-rate interactions. Measurement-noise adaptation is separated from model-parameter updating: a relative-volatility-derived time-varying factor adjusts the measurement-noise covariance and detects abrupt operating-condition changes, thereby suppressing the effects of capacity regeneration and local anomalies on state estimation. Model parameters and the Wiener diffusion intensity are updated by a sliding-window Bayesian estimator using maximum likelihood estimation (MLE) priors, time-decay weights, and filtering-uncertainty weights; the resulting process-noise covariance is fed back to the UKF time prediction at the next step. Validation on the complementary CALCE, NASA PCoE, and Wenzhou Randomized datasets shows that the proposed method improves degradation-modeling accuracy under controlled stress, the robustness of online estimation under nonstationary noise, and the generalizability of RUL predictions under dynamic operating conditions.

The advances in technology in the manufacturing industry has enabled the construction of intricate systems comprising multiple key components. The degradation of these components can usually be characterized based on multiple indicators. The correlation between multiple indicators of any component and the interdependence between components cannot be ignored. Moreover, time-varying loads and dynamic conditions exert additional random effects on the system degradation process, which affects the remaining useful life (RUL) prediction. Thus, in an attempt to overcome these issues, this study proposes a hierarchical system RUL prediction method under the influence of the external environment that considers two different kinds of the random correlations among multiple indicators of any component and among multiple components. First, multiple indicators representing the degradation of any component are constructed using multiple features. A distributed fusion degradation modeling is performed by considering the stochastic correlation among the indicators of a component. Second, considering the stochastic relevance among multiple components and the influence of measurable and unmeasurable external environments, the degradation model of the multi-component system is constructed. Meanwhile, the parameters in the model are estimated and updated step-by-step. Furthermore, the RUL of a multi-component system with different structural types is predicted. Finally, the effectiveness of the proposed method is verified based on numerical examples and practical cases. Compared with the existing methods, the root mean square error of this method is reduced by 17%, and the SF is increased by 19%. This model can more intuitively reflect the hierarchical random correlation between complex multi-component system degradations.

To meet the requirements of a new energy power system for accommodation, coal-fired units need to frequently and rapidly change their load, and precise and rapid estimation of their operating state is crucial for ensuring safety and control performance. However, owing to the nonlinear and strongly coupled characteristics of coal-fired units, existing state estimation methods (such as Kalman filters) have difficulty meeting the new requirements of coal-fired units. Therefore, this paper proposes a novel digital twin modeling method based on a lightweight denoising diffusion probabilistic model (CFU-StateDiff) for state estimation in coal-fired units. First, an integration of expert knowledge and a data-driven approach for state variables (IEDA-SV) is proposed to construct a high-quality state estimation dataset. Second, MobileNet’s depthwise separable convolution is adopted to reduce the model’s parameter count, and the Mish function is introduced to retain negative state information under nondesign conditions and ensure the completeness of information across all operating conditions. Third, a joint optimization strategy combining feature guidance and physical constraints is employed, in which expert features are embedded as conditions into the diffusion sampling process and a physical consistency loss function based on the law of conservation of mass is introduced to ensure that the generated state variables comply with the mass conservation relationship of the working medium. Experimental engineering results reveal that CFU-StateDiff balances rapid state estimation with physical consistency, achieving a 4.38% improvement in the latter over the optimal baseline conditional denoising diffusion probabilistic model (CDDPM).

Remaining useful life (RUL) prediction is critical for industrial equipment stability and health management. To comprehensively characterize system degradation and precisely predict the system's RUL, the impact of dynamic environment on degraded systems must be considered. Therefore, a method is proposed in this study that integrates dynamic environmental effects and the joint correlation between degradation rate and volatility into a nonlinear Wiener process for degradation modeling. In addition, to fully reflect the system health state, a state-space model for the degraded system is established by this study with consideration of multiple degradation indicators and their correlations. Utilizing stochastic correlation properties, RUL prediction models for systems across various failure modes are developed. To address the degradation state and parameter updating issues in the established complex degradation model, an adaptive kernel density estimation auxiliary particle filter algorithm is proposed. This method enhances the performance of particle filtering through adaptive bandwidth adjustment and updates particle weights according to observation data during resampling, enabling real time updating of both the degradation state and model parameters. Finally, the effectiveness of the proposed method is validated through simulation experiment and the dataset XJTU-SY.