The reliability of hydraulic valves is critical for the safety and efficiency of industrial systems. While vibration and pressure sensors are widely deployed for condition monitoring, leveraging the heterogeneous data from these multi-sensor systems for accurate remaining useful life (RUL) prediction remains challenging due to noise, outliers, and inconsistent sampling rates. This study proposes a sensor data-driven framework that integrates multi-step signal preprocessing, time–frequency feature fusion, and a machine learning model to address these challenges. Specifically, raw data from vibration and pressure sensors are first harmonized through a multi-step preprocessing pipeline including Hampel filtering for impulse noise, Robust Scaler for outlier mitigation, Butterworth low-pass filtering for effective frequency band retention, and resampling to a unified rate. Subsequently, vibro-acoustic features are extracted from the preprocessed sensor signals, including Fast Fourier Transform (FFT)-based frequency domain features and Wavelet Packet Decomposition (WPD)-based time–frequency features, to comprehensively characterize the valve’s degradation. A health indicator (HI) is constructed by fusing the most sensitive features. Finally, a Kernel Principal Component Analysis (KPCA)-optimized Random Forest model is developed for HI prediction, which strongly correlates with RUL. Validated on the UCI hydraulic condition monitoring dataset through 20-run Monte-Carlo cross-validation, our method achieves a root mean square error (RMSE) of 0.0319 ± 0.0090, a mean absolute error (MAE) of 0.0109 ± 0.0014, and a coefficient of determination (R²) of 0.9828 ± 0.0097, demonstrating consistent performance across different data partitions. These results confirm the framework’s effectiveness in translating multi-sensor data into actionable insights for predictive maintenance, offering a viable solution for industrial health management systems.

Natural disasters can split distribution networks into isolated microgrids, and the remaining local storage is often not enough to keep critical loads supplied until repairs start. Mobile emergency power vehicles (MEPVs) are useful in bridging this gap, but their dispatch is easily distorted when travel time is treated as static and battery charging is treated as constant-rate. This paper studies MEPV routing and load restoration under scenario-based road degradation and non-linear constant-current/constant-voltage (CC-CV) charging. A limited-communication setting is also stated explicitly, since post-disaster information is usually intermittent rather than fully real-time. The resulting scheduling problem is solved by an Improved Adaptive Harmony Search (IAHS) algorithm. IAHS keeps the harmony–memory structure but moves binary load decisions through a continuous latent space before threshold decoding, so that the differential update can use population differences without repeatedly disturbing load variables that have already become stable. Tests on a six-area system show higher average recovery and lower dispersion than HS, AHS, GHS, and AHS with standard BDE. Additional road-degradation sensitivity tests and single-MEPV scale-up tests show that the method remains feasible under stronger road deterioration and that computation time grows in a manageable way as the number of load variables increases.

Extreme disasters can simultaneously damage distribution equipment and communication links, preventing control centers from promptly acquiring complete energization information for critical nodes. This study formulates joint critical-node state identification in a fixed post-disaster radial topology as exact maximum a posteriori (MAP) inference in a tree-structured Bayesian network and proposes a path-cached branch-and-bound search (PCBS). PCBS combines admissible upper bounds for partial assignments, posterior-state discriminability ordering, bounded-width initial candidate construction, branch-and-bound pruning, and rollback-enabled path caching without changing the probabilistic model, candidate space, or MAP objective. Across 200 random scenarios on the IEEE 33-bus system, PCBS returned exactly the same MAP solutions and joint probabilities as exhaustive enumeration. On the IEEE 85-bus system with 14 critical nodes, the runtime speedup reached 28.3. When the evidence-quality score increased from 0.24 to 0.82, state accuracy rose from 78.30% to 92.40%, while runtime decreased from 0.714 s to 0.513 s. On the 300-node radial network, PCBS completed inference in 1.13 s on average with 30 critical nodes and 2.46 s with 60 critical nodes, while maintaining approximately 86.5–86.7% state accuracy. Hardware-in-the-loop experiments on a shipboard power system achieved 85.00% mean accuracy with a 1.31 s mean end-to-end inference time.