The high operational cost of aircraft, limited availability of air space, and strict safety regulations make training of fighter pilots increasingly challenging. By integrating Live, Virtual, and Constructive simulation resources, efficiency and effectiveness can be improved. In particular, if constructive simulations, which provide synthetic agents operating synthetic vehicles, were used to a higher degree, complex training scenarios could be realised at low cost, the need for support personnel could be reduced, and training availability could be improved. In this work, inspired by the recent improvements of techniques for artificial intelligence, we take a user perspective and investigate how intelligent, learning agents could help build future training systems. Through a domain analysis, a user study, and practical experiments, we identify important agent capabilities and characteristics, and then discuss design approaches and solution concepts for training systems to utilise learning agents for improved training value.

In this paper, we expolore Multi-Agent Reinforcement Learning (MARL) methods for unmanned aerial vehicle (UAV) cluster. Considering that the current UAV cluster is still in the program control stage, the fully autonomous and intelligent cooperative combat has not been realised. In order to realise the autonomous planning of the UAV cluster according to the changing environment and cooperate with each other to complete the combat goal, we propose a new MARL framework. It adopts the policy of centralised training with decentralised execution, and uses Actor-Critic network to select the execution action and then to make the corresponding evaluation. The new algorithm makes three key improvements on the basis of Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm. The first is to improve learning framework; it makes the calculated Q value more accurate. The second is to add collision avoidance setting, which can increase the operational safety factor. And the third is to adjust reward mechanism; it can effectively improve the cluster’s cooperative ability. Then the improved MADDPG algorithm is tested by performing two conventional combat missions. The simulation results show that the learning efficiency is obviously improved, and the operational safety factor is further increased compared with the previous algorithm.

Optimisation-based control design techniques are applied to multicopters with variable-RPM rotors. The handling qualities and motor current requirements of a quadcopter, hexacopter and octocopter with equal gross weights (5,360N) and total disk areas (producing a 287N/m $^2$ disk loading) are compared in hover. For axes that rely on the rotor thrust (all except yaw), the increased inertia of the larger rotors on the quadcopter increase the current requirement, relative to vehicles with fewer, smaller rotors. Both the quadcopter and hexacopter have maximum current margin requirements (relative to hover) during a step command in longitudinal velocity. In yaw, rotor inertia is irrelevant, as the reaction torque of the motor is the same whether the rotor is accelerating or overcoming drag. This, combined with the octocopter’s greater inertia as well as the fact that it requires 30% less current to drive its motors in hover, results in the octocopter requiring the greatest current margin, relative to hover conditions. To meet handling qualities requirements, the total weight of the motors of the octocopter and hexacopter is comparable at 13.5% weight fraction, but the quadcopter’s motors are heavier, requiring 16% weight fraction. If the longitudinal and lateral axes were flown in ACAH mode, rather than TRC mode, the total motor weight of all configurations would be nearly identical, requiring about 13.5% weight fraction for motors (compared to 7–9% weight fraction from hover torque requirements).

This paper investigates the feasibility of improving the aircraft landing performance by design the damping orifice parameters of the landing gear using lattice Boltzmann method coupled with the response surface method. The LBM is utilised to simulate characteristics of the damping orifice after model validation. The numerical model of the landing gear using simulated damping force is validated by single landing gear drop test. Based on the numerical model and the response surface functions, the sensitivity analysis and the optimisation design are performed. The maximum error of mean velocity simulated using LBM with experimental data is 7.07% for sharp-edged orifices. Moreover, the numerical model predicts the landing responses adequately, the maximum error with drop test data is 2.51%. The max overloading of the aircraft decreases by 5.44% after optimisation, which proves that this method is feasible to design the damping orifice for good landing performance.

The aim of this paper is to provide preliminary results on a traffic coordination framework based on stochastic task allocation. General trends and the predicted advent of personal aerial vehicles increase traffic rapidly, but current air traffic management methods admittedly cannot scale appropriately. A hierarchical system is proposed to overcome the problem, the middle layer of which is elaborated in this paper. This layer aims to enable stochastic control of traffic behaviour using a single parameter, which is achieved by applying distributed stochastic task allocation. The task allocation algorithm is used to allocate speeds to vehicles in a scalable way. By regulating the speed distribution of vehicles the conflict rates remain manageable. Multi-agent simulation results show that it is possible to control ensemble dynamics and together with that traffic safety and throughput via a single parameter. Using transient simulations the dynamic performance of the system is analysed. It is shown that the traffic conflict reduction problem can be transformed into a control design problem. The performance of a simple controller is also evaluated. It was shown that by applying the controller, quicker transients can be achieved for the mean speed of the system.

The cooperative guidance problem of multiple inferior missiles intercepting a hypersonic target with the specific impact angle constraint in the two-dimensional plane is addressed in this paper, taking into consideration variations in a missile’s speed. The guidance law is designed with two subsystems: the direction of line-of-sight (LOS) and the direction of normal to LOS. In the direction of LOS, by applying the algebraic graph theory and the consensus theory, the guidance command is designed to make the system convergent in a finite time to satisfy the goal of cooperative interception. In the direction of normal to LOS, the impact angle is constrained to transform into the LOS angle at the time of interception. In view of the difficulty of measuring unknown target acceleration information in real scenarios, the guidance command is designed by utilising a super-twisting algorithm based on a nonsingular fast-terminal sliding mode (NFTSM) surface. Numerical simulation results manifest that the proposed guidance law performs efficiently and the guidance commands are free of chattering. In addition, the overall performance of this guidance law is assessed with Monte Carlo runs in the presence of measurement errors. The simulation results demonstrate that the robustness can be guaranteed, and that overall efficiency and accuracy in intercepting the hypersonic target are achieved.

This paper presents the extension and validation of omni-failure envelopes for first-ply failure (FPF) and last-ply failure (LPF) analysis of advanced composite materials under general three-dimensional (3D) stress states. Phenomenological failure criteria based on invariant structural tensors are implemented to address failure events in multidirectional laminates using the “omni strain failure envelope” concept. This concept enables the generation of safe predictions of FPF and LPF of composite laminates, providing reliable and fast laminate failure indications that can be particularly useful as a design tool for conceptual and preliminary design of composite structures. The proposed extended omni strain failure envelopes allow not only identification of the controlling plies for FPF and LPF, but also of the controlling failure modes. FPF/LPF surfaces for general 3D stress states can be obtained using only the material properties extracted from the unidirectional (UD) material, and can predict membrane FPF or LPF of any laminate independently of lay-up, while considering the effect of out-of-plane stresses. The predictions of the LPF envelopes and surfaces are compared with experimental data on multidirectional laminates from the first and second World-Wide Failure Exercise (WWFE), showing a satisfactory agreement and validating the conservative character of omni-failure envelopes also in the presence of high levels of triaxiality.

Weapon target allocation (WTA) is an effective method to solve the battlefield fire optimisation problem, which plays an important role in intelligent automated decision-making. We researched the multitarget allocation problem to maximise the attack effectiveness when multiple interceptors cooperatively attack multiple ground targets. Firstly, an effective and reasonable fitness function is established, based on the situation between the interceptors and targets, by comprehensively considering the relative range, relative angle, speed, capture probability and radiation source matching performance and thoroughly evaluating them based on the advantage of the attack effectiveness. Secondly, the optimisation performance of the particle swarm optimisation (PSO) algorithm is adaptively improved. We propose an adaptive simulated annealing-particle swarm optimisation (SA-PSO) algorithm by introducing the simulated annealing algorithm into the adaptive PSO algorithm. The proposed algorithm can enhance the convergence speed and overcome the disadvantage of the PSO algorithm easily falling into a local extreme point. Finally, a simulation example is performed in a scenario where ten interceptors cooperate to attack eight ground targets; comparative experiments are conducted between the adaptive SA-PSO algorithm and PSO algorithm. The simulation results indicate that the proposed adaptive SA-PSO algorithm demonstrates great performance in convergence speed and global optimisation capabilities, and a maximised attack effectiveness can be guaranteed.