Haozhong Huang

High-power linear alternators (HPLAs) are key energy conversion components in free-piston Stirling generators (FPSGs), yet their output performance strongly depends on the coupled matching among the gas pressure wave, piston dynamics, and external electrical load. In this work, the electrical load-matching characteristics and loss mechanisms of a gas-coupled HPLA were investigated experimentally and numerically. A thermodynamic–dynamic–electromagnetic coupling model, validated against gas-coupled dynamic experiments, is established to account for the position-dependent electromagnetic parameters and the coupled gas mechanical–electrical interaction. The results show that load resistance mainly governs the current level and the loss composition, whereas series capacitance shifts the electrical phase and modifies the effective electromagnetic stiffness and damping acting on the piston. Consequently, the maximum power condition does not coincide with complete electrical resonance: allowing the capacitance to deviate from resonance raises the maximum output power from 13.54 kW to 14.58 kW, a gain of about 7.7%. Moreover, as the load resistance increases, the dominant loss migrates from current-dependent electromagnetic loss to mechanical damping loss, so that the maximum power point and the maximum efficiency point are clearly separated. These findings establish a coupled load-matching framework for selecting the operating point and external load according to design priority, rather than by electrical resonance alone.

The energy management strategy critically determines hybrid electric vehicle (HEV) overall performance. To fully exploit cooptimization potential between energy allocation and thermal management, this article proposes a reinforcement learning (RL) model based on joint cross-domain data feature learning. The core model is further enhanced by integrating a bidirectional long short-term memory network. This enhancement enables the RL agent to jointly perceive and adapt to multi-condition driving patterns for globally optimized control decisions. For evaluation, a coupled HEV powertrain-thermal management model is developed and validated using multiple standard driving cycles. Experimental results show that, compared to existing mainstream RL methods, the proposed approach demonstrates superior performance. It more effectively maintains battery state-of-charge stability, keeps battery and motor temperatures within safe limits, and significantly reduces equivalent fuel consumption per 100 km. Overall, the method exhibits strong robustness and application potential, offering an effective solution for HEV energy management under complex driving conditions.

To solve the problems of poor fuel atomization effect, low combustion efficiency, and uneven temperature distribution of the evaporator tube of a certain micro turbojet engine, a structural optimization design method based on a genetic algorithm is proposed. Taking the inner diameter of the evaporator tube, the diameter of the nozzle hole, the number of nozzle holes as design variables, the fuel atomization particle size (d₅₀), combustion efficiency (η), and maximum wall temperature (T_max) as optimization objectives, a multi-objective optimization mathematical model is established. The iterative optimization is carried out through the selection, crossover, and mutation operations of the genetic algorithm, and the optimization effect is verified by combining CFD (Computational Fluid Dynamics) numerical simulation. The results show that when the inner diameter of the evaporator tube is 2.6 mm, the diameter of the nozzle hole is 0.8 mm and the number of nozzle holes is eight, the fuel atomization particle size of the evaporator tube is reduced by 18.3%, the combustion efficiency is increased by 7.6%, and the maximum wall temperature is decreased by 12.4%, which significantly improves the working performance of the evaporator tube and provides an effective reference for the optimization design of key components of micro turbojet engines.

Accurate prediction of voltage degradation in proton exchange membrane fuel cells (PEMFCs) is crucial for ensuring their reliability and enabling predictive management. While existing methods predominantly focus on point prediction of voltage, the quantification of predictive uncertainty has not been adequately addressed. To tackle this issue, an LP-TimesNet model is proposed as a novel approach for probabilistic prediction of voltage degradation. This model integrates the TimesNet module to capture multiperiod temporal dependencies and incorporates a Laplace distribution-based output layer to quantify predictive uncertainty. Extensive evaluations under dynamic conditions demonstrate that the proposed model achieves a 19.4% reduction in root mean square error (RMSE) compared to the baseline TimesNet model, along with sharper and more reliable confidence intervals than other probabilistic models. Under both static and quasi-static conditions, the model also exhibits improved predictive accuracy and better uncertainty calibration. These results indicate that the proposed model possesses strong generalization capabilities, yielding accurate voltage predictions and reliable interval estimates across diverse operating conditions.

Rotational speed is a key performance parameter of aero-engines. As an indispensable parameter in control systems, rotational speed has received widespread attention. However, research on rotational speed control methods for micro-turbojet engines used in unmanned aerial vehicles is relatively limited. Therefore, this paper proposes a rotational speed controller for micro-turbojet engines based on the Improved Spider Wasp Optimization (ISWO) algorithm, which adopts Linear Active Disturbance Rejection Control (LADRC) technology. Data from the actual operation of the micro-turbojet engine were collected, and the engine rotational speed models under three operating conditions were obtained through system identification. The proposed controller was applied to these rotational speed models, along with the classic Proportional–Integral–Derivative (PID) controller, the LADRC controller, and the LADRC controller based on the original Spider Wasp Optimization (SWO) algorithm. Simulation experiments and comparisons of the controllers were conducted based on performance indicators. The simulation results show that under the three operating conditions, the ISWO-LADRC controller outperforms the other three controllers in indicators such as rise time, overshoot, and settling time, all of which are smaller. Thus, ISWO-LADRC exhibits better comprehensive control performance.