With the rapid advancement of the Internet of Things and edge intelligent computing, the demand for low-power, high-energy-efficient computing systems has become increasingly urgent. The traditional von Neumann architecture suffers from poor energy efficiency in data-intensive tasks due to the 'memory wall' problem. Neuromorphic computing, as an emerging paradigm that mimics the biological brain's information-processing methods, offers highly promising solutions to overcome energy-efficiency bottlenecks through event-driven operations, integrated sensing, storage, and computation, and novel devices such as memristors. This paper systematically analyses and summarises the latest research achievements in neuromorphic computing across hardware devices, system architectures, and optimisation strategies. It first presents fundamental hardware implementation options for simulating neurons and synapses, highlighting memristors' benefits in low-power synaptic plasticity. Then the system-level low-power architectures are discussed-including the event-driven paradigm and compute-in-memory concept integration. Finally, taking IoT edge nodes as an example application scenario of energy-efficient neuromorphic systems, it also outlines current major issues confronting technologies along with future development directions.

The accurate prediction of power losses in ultra-high-voltage (UHV) AC transmission systems is critical for ensuring the economic and efficient operation of modern power grids. This paper investigates the temperature-dependent fluctuations in line losses—particularly those induced by corona discharge—under varying ambient conditions. To this end, this paper proposes a comprehensive loss-prediction framework integrating Spearman rank correlation analysis with a particle swarm optimization–enhanced extra-trees (PSO-ET) model. The methodology is validated using operational data from 1000 kV UHV transmission projects in the Fujian–Zhejiang region. First, Spearman correlation analysis confirms temperature as the dominant meteorological factor influencing line losses. Subsequently, the PSO algorithm is employed to globally optimize three key hyperparameters of the ET model: tree count, maximum tree depth, and the humidity threshold used to dichotomize operating conditions into "dry" and "high-humidity" regimes. This enables precise decoupled training under distinct environmental states. Experimental results demonstrate that the proposed model achieves an average absolute error (MAE) of only 5.1315 MW on the independent test set. Further sensitivity analysis reveals that the gradient of line loss with respect to temperature peaks at approximately 7.6 °C—indicating a pronounced nonlinear response. Crucially, this inflection coincides with the onset of surface condensation, thereby uncovering a previously underappreciated surge effect: in low-temperature, high-humidity environments, condensation on conductor surfaces significantly amplifies corona losses. This study provides a robust, data-driven foundation for real-time loss monitoring and energy-efficient scheduling of UHV transmission lines under complex and dynamic meteorological conditions.

The aim of this study is to examine the aerodynamic characteristics of a 2018 Formula 1 car rear wing, which is equipped with a Drag Reduction System (DRS), by applying a two-dimensional computational fluid dynamics (CFD) approach, particularly when the DRS is partially activated. The study applied a commercial CFD code, ANSYS Fluent, based on a k-omega Shear Stress Transport (SST) turbulence closure model, to solve the flow problem. The CFD code was applied to a 2018 Formula 1 car rear wing to solve the aerodynamic problem for different DRS activation ratios, i.e., 0%, 25%, 50%, 75%, and 100%, at three different free-stream velocities, i.e., 50 m/s, 70 m/s, and 90 m/s. The results obtained in this study indicated that although aerodynamic forces are proportional to the square of the free-stream velocity, lift and drag coefficients remain almost constant for a wide range of free-stream velocities. The results reveal a "sweet spot" where a 25% opening of the DRS resulted in a 50% reduction in drag and a 20-22% increase in downforce compared to a fully closed DRS position due to a "slot effect" delaying the flow separation. The results obtained in this study revealed that the highest opening of the DRS, i.e., 100% DRS, resulted in a 75% reduction in drag but caused a significant reduction in downforce.

The performance of robot control systems directly influences their effectiveness across various industries, agriculture, services, and other sectors. Traditional PID control is widely used due to its simple structure and ease of implementation. However, when facing complex nonlinear systems, time-varying parameters, and external disturbances, its control accuracy and robustness significantly decrease. This paper analyzes the limitations faced by classical PID control in robot applications, with a focus on exploring how fuzzy PID control can adaptively adjust PID parameters by introducing fuzzy inference mechanisms, thereby improving the system's adaptability in uncertain environments. Furthermore, it reviews the integration strategies of various intelligent optimization algorithms, modern control methods, and PID control, including particle swarm optimization (PSO), genetic algorithm (GA), active disturbance rejection control (ADRC), and PID composite control. The results demonstrate the significant effects of these methods in improving robot trajectory tracking accuracy, vibration suppression, and anti-interference ability. It further found that by integrating intelligent algorithms with advanced control strategies, the performance of PID controllers can be systematically improved, providing an effective technical path for the reliable operation of robot control systems in complex dynamic environments.