Motor servo systems in practical applications often face input delay and inherent model uncertainties, posing significant challenges to control design. To enhance the tracking accuracy of motor servo systems with input delay, this paper proposes a novel neural network-based adaptive control method with prescribed performance guarantees. Firstly, a mathematical model of motor servo system incorporating input delay is established. Subsequently, as the core of the proposed control methodology, the auxiliary dynamic related to input delay is defined, neural network technique is employed to approximate both the unknown system dynamic and input delay-related quantity, while a barrier function integrating prescribed performance is introduced to generate designed error signals. Then the controller is systematically derived through this framework. Through Lyapunov stability analysis, the proposed controller is rigorously proven to compensate for input delay while achieving prescribed performance. Comparative experiments validate the effectiveness of the proposed control strategy.
Based on the issues of substantial torque fluctuation, severe magnetic leakage, and low efficiency in conventional rectangular embedded permanent magnet synchronous motors (PMSM), this paper introduces an arc-trapezoidal magnet (ATM) structure aimed at enhancing the torque performance of the motor. Through comprehensive simulation analysis and experimental validation, the design parameters are optimised via orthogonal experimental, leading to the identification of the optimal parameter set. Experimental results reveal a 26.8% increase in average motor torque, a 51.2% reduction in torque fluctuation, and a significant improvement in the motor's output torque performance.
This study investigates the nonlinear thermo-mechanical behaviour of functionally graded (FG) hyperbolic rotating disks with rigid shaft inclusions and radially varying density. The disks are composed of Ti-6Al-4V/GFRP composites, with properties following a power-law gradation indexed by m. Using a generalised strain-displacement formulation, non-dimensional equilibrium equations capture axisymmetric deformations under combined thermal and mechanical loading. Numerical results show that material gradation and shaft geometry significantly influence performance. Under isothermal conditions, GFRP-rich disks (m = 0) achieve up to 42% higher yielding angular speed than Ti-6Al-4V-rich disks (m =1.25), while thermal loading reduces this advantage. Radial stress at the bore decreases similar to 22% in GFRP-rich disks, whereas circumferential stress rises 15%-20% in Ti-6Al-4V-rich disks. Radial displacements increase 27%-30%, further amplified 16%-28% under thermal effects. These findings highlight the trade-off between high-speed capacity and thermo-mechanical stability, providing a framework for optimising FG disks in aerospace, energy storage, and hydromechatronic systems.
Water-jet-guided laser (WJGL) systems require ultra-stable hydraulic pressure (<5% pulsation amplitude at 0-1000 Hz). Current viscoelastic-lined pulsation attenuators lack standardised characterisation methods, leading to significant errors in impedance modelling. This study proposes a novel end-correction formula incorporating viscoelastic material parameters (sound speed, bulk modulus) for extended-tube expansion chambers (VMETC). Based on the finite element analysis and experimental validation results, the formula demonstrates good universality across 441 parameter combinations, with 71.4% of the combinations exhibiting errors below 5%.
The convergence of large language models (LLMs), structured knowledge bases (KBs), and reasoning ability (RA) presents a promising trajectory toward general embodied intelligence (GEI). This paper reviews the evolution of LLM-centered intelligent systems, emphasising their integration with knowledge representation, logical reasoning, and physical embodiment. We analyse LLM architectures, pre-training methods, and inference mechanisms, along with their interaction with external knowledge sources and structured reasoning frameworks. Furthermore, we examine embodied intelligence (EI) paradigms wherein agents learn and act in physical environments. To synthesise these dimensions, we present a conceptual framework that illustrates the synergy among LLMs, KBs, RA, and embodiment, serving as a guiding model for perception, reasoning, and action rather than an implemented engineering architecture. To advance toward GEI, we identify five key challenges: efficient LLM deployment, closed-loop knowledge integration, hybrid symbolic-neural reasoning, perception-action grounding, and continual learning. This survey provides a comprehensive roadmap for developing adaptive, multimodal agents capable of operating in complex, dynamic settings.
In ionic liquid hydrogen compressors, the liquid provides functions of sealing, piston lubrication, and hydrogen cooling. However, some liquids are expelled with hydrogen during the discharge, leading to a gradual decrease in liquid volume. Liquid replenishment is therefore necessary to maintain an adequate liquid level for effective sealing and lubrication, while also enhancing cooling efficiency and assisting the compression process closer to isothermal conditions. In this study, a transient two-phase flow model was developed to evaluate the thermal performance of the compressor under three different liquid replenishment schemes. The volume of fluid method was employed to track the interface between hydrogen and ionic liquids. Meanwhile, the two-phase flow and heat transfer were simulated over a complete compression cycle. Results showed that replenishment improved compression performance, achieving a minimum polytropic index of 1.145 and a maximum isothermal efficiency of 86.06%. However, replenishment occupied part of the suction stroke and increased liquid discharge, reducing volumetric efficiency, with the lowest value of 69.23%. Among the three schemes, the cylinder-top replenishment method demonstrated an outstanding trade-off between isothermal efficiency (83.51%) and volumetric efficiency (89.92%), making it a promising strategy for practical hydrogen refuelling applications.
This paper addresses nonlinear vibrations and low energy efficiency in deep well drilling in complex geological settings. A novel electromechanical-hydraulic-gas (EHG) coupled rotary punch drilling (RPD) system is investigated. Firstly, dynamic characteristics of key components are modelled, establishing a foundation for system modelling. Subsequently, a multi-physics EHG simulation model for the RPD system is established using AMESim and Simulink for co-simulation, thoroughly validating its accuracy and stability under transient loads. Results show that with increased impact load amplitudes, pressure response times in feed and rotary systems remain below 1.5 s, flow fluctuation under 3%, and drill string velocity stabilises within 2 s. High-frequency impacts produce synchronous pressure and flow adaptations. The pneumatic impact mechanism reaches a peak piston velocity of 62.84 m/s, maintaining effective rock-bit contact. These results further enhance drilling rig impact resistance, energy efficiency, and control of dynamic characteristics.
To address the time-varying output force and compliance issues in pneumatic artificial muscle-driven robots, this paper proposes a bionic ankle joint inspired by human muscle structure. By recruiting multiple pneumatic artificial muscles and implementing compliance control, the overall output performance is enhanced. This study presents the structural design of the bionic ankle joint with multiple pneumatic artificial muscles and analyses its bionic actuation mechanism. Furthermore, impedance control and adaptive impedance control models are developed, followed by a simulation-based analysis of the output angular characteristics and force response. A performance evaluation platform is constructed for experimental validation. The results indicate that the bionic ankle joint with adaptive impedance control effectively reduces angular tracking errors and output force deviations, while improving joint compliance.
This study introduces a cost-effective framework to optimise asphalt crack sealing by converting binary crack images into an open-loop travelling salesman problem, avoiding the need for an accurate crack skeleton. To address sub-optimality, we propose an improved discrete grey wolf optimiser (I-DGWO) that integrates reverse sequence shift mutation and a 'ruin and recreate' strategy to enhance exploration and exploitation. In addition, we incorporate the Lin-Kernighan-Helsgaun-3 (LKH-3) as an initialisation technique to search around the solution. Our statistical evaluation shows a substantial reduction in overlapping distance of up to 27% and 63% on single and multiple cracks respectively when compared to four other baselines. Furthermore, the ablation studies show that the addition of the 'ruin and recreate' reduces the overlapping distance by 12% and the incorporation of the reverse sequence shift mutation further reduces the overlapping distance by 21%.
