Eng

Underwater acoustic sensor networks (UASNs) are severely affected by propagation delay, limited bandwidth, and constrained node energy. This work presents EXTREPSO, a machine learning framework that combines an Extra Trees Regressor with Particle Swarm Optimization (PSO) for offline hyperparameter tuning of a hop-level energy-efficiency predictor. The target is derived from Aqua-Sim NG trace fields using the payload size, the simulator Tx_power field, and observed hop delay. Aqua-Sim NG defines PT as transmitted signal power in watts, so the target has dimensional units of bit/J; however, because the observed hop delay includes propagation and protocol-related waiting, it is treated as an operational trace-derived metric rather than a state-resolved modem-energy measurement. In the original 80:20 benchmark, Extra Trees was the strongest baseline among seven classical ML and three DL models. A leakage-resistant sender-grouped outer five-fold evaluation with inner grouped validation was subsequently used to compare baseline Extra Trees, PSO-tuned Extra Trees, and an interpretable inverse-power baseline on untouched test folds. PSO tuning produced a modest average improvement over baseline Extra Trees (mean RMSE 2107.31 versus 2230.58), but the improvement was not consistent across folds; the inverse-power baseline achieved the lowest mean RMSE (1765.88). A separate five-run equal-budget PSO/GWO sensitivity check also found no meaningful superiority of PSO over GWO. These results support the trace-driven prediction and hyperparameter-tuning workflow while limiting claims of universal optimizer benefit or direct network-energy improvement.

Accurate short-term load and photovoltaic (PV) power forecasting is essential for power-system dispatch and renewable-energy accommodation under high PV penetration. However, measured source-load data often contain outliers and missing values, while the forecasting performance of bidirectional long short-term memory (BiLSTM) networks is sensitive to hyperparameter selection. This paper develops an integrated short-term load and PV power forecasting framework that combines characteristic-aware robust data preprocessing with BiLSTM modeling tuned by quantum particle swarm optimization (QPSO). An iterative 3σ criterion is first used to detect spike, drop, negative, and flat anomalies according to the different characteristics of load and PV data, and piecewise cubic Hermite interpolation is then applied to repair abnormal and missing points. A BiLSTM network is used to capture bidirectional temporal dependencies, and QPSO is introduced to optimize the number of hidden units, maximum training epochs, and learning rate. Experiments were conducted on measured load and PV datasets with a 15 min resolution, and the proposed model was compared with long short-term memory (LSTM), BiLSTM, and a BiLSTM tuned by particle swarm optimization (PSO), denoted PSO-BiLSTM. In the primary forecasting evaluation on measured grid data, QPSO-BiLSTM achieved mean absolute percentage error (MAPE) values of 2.1616% for load forecasting and 7.5687% for PV power forecasting. Additional random-seed-controlled ablation and robustness validation confirmed statistically significant contributions of the preprocessing procedure and QPSO to load forecasting, while the proposed framework maintained competitive performance for PV forecasting. The proposed method improves forecasting accuracy and provides reliable data support for short-term operation and dispatch of renewable-energy power systems.

This paper studies the stationary balancing of a full-scale bicycle robot using a scissored-pair control moment gyroscope (CMG). The test robot weighs 60 kg and includes two counter-rotating flywheels, a 20:1 geared gimbal system, an inertial measurement unit (IMU), and an embedded controller. For controller design, a simplified nonlinear model of the roll and CMG dynamics is derived and linearized around the upright position. The main contribution of this work is the experimental testing of a PDD2-LQR controller. This controller keeps the standard LQR state-feedback baseline and adds feedback from the roll error, roll rate, and filtered roll acceleration. Instead of redesigning the whole controller, the same LQR gains are used in both the baseline LQR and PDD2-LQR setups. This allows us to directly evaluate the benefit of the extra PDD2 terms on the same hardware under the same actuator limits. Experiments were carried out using PID, LQR, and PDD2-LQR controllers. The PDD2-LQR controller reached an average RMSE of 0.660 degrees and an average MAE of 0.477 degrees. These values represent improvements of about 51.8% and 43.2%, respectively, compared with the standard LQR controller. Overall, the results show that adding roll-motion feedback to the baseline controller improves the stationary balancing performance under the tested conditions.