To investigate the effect of ice accretion on helicopter rotors in forward flight in different flight states, a helicopter rotor icing test system was developed. A rotor model icing test was conducted in the 4.8 m× 3.2 m test section of the China Aerodynamics Research and Development Center icing wind tunnel. The effect of the rotational speed and initial thrust on the performance of a 2 m-diamter rotor model was examined. During the icing test, the control angle of the rotor was maintained constant, and the dynamic variation of the rotor thrust and torque was measured by a balance. The ice shape of the airfoil at typical span stations and of the entire blade was measured by two-dimensional ice cutting and three-dimensional scanning system, respectively. Test results showed that ice accretion occurred primarily on the blade leading edge and lower surface, and that icing increased rotor torque and power while decreasing rotor thrust. The rod end bearing of the pitch link could be stuck by accreted ice, resulting in a loss of rotor control.

To train pilots' short-range air combat skills, the traditional way based on flight simulation technology is to have multiple pilots operate multiple fighter simulators at the same time. If an intelligent virtual opponent is used to assist pilots in confrontation training, not only could the normal training process without other pilots be guaranteed, but the training cost could also be reduced to a great extent. In this paper, an integrated method based on dynamic Bayesian network (DBN) and constrained gradient algorithm is proposed to realize autonomous decision making and space occupancy guidance for intelligent virtual opponents in the attack and defense game during short-range air combat training. A dynamic Bayesian network model for short-range air combat decision making is established in combination with the space occupying situation, the fire control attack area and the identification results of maneuvering actions. This model realizes an intelligent selection of occupancy guidance index in accordance with the dynamic battlefield environment. A target trajectory prediction model is built for each type of maneuvers identified online to obtain the real-time prediction of the target trajectory. With the occupancy guidance index, target trajectory predication, and the flight performance constraints in consideration, a constraint gradient method is used to calculate the optimal occupancy guidance quantity of the intelligent virtual opponent. Thus, a seamless combination of space occupancy decision and guidance quantity calculation for intelligent virtual opponent is achieved. The simulation results of short-range air combat show that the proposed method can realize rational autonomous decision making and space occupancy guidance for intelligent virtual opponent, overcome the problem of solidifying the maneuver mode in traditional methods, and thus have better real time and optimization performance.

The additive manufacturing (AM) technology develops rapidly and it is widely employed in the fabrication of aerospace alloy components.Many additive manufacturing alloy components bear cyclic loadings, and the fatigue failure is very common. The fatigue damage model considering the influence of additive manufacturing process is established and the fatigue lives of additive manufacturing alloy materials are predicted. The elastic-plastic constitutive model and the fatigue damage model considering the additive manufacturing process parameters are presented, and the finite element numerical method is presented for the fatigue life computation. The fatigue lives of additive manufacturing metal materials are predicted, which are basically consistent with the experimental results, and the computed errors are analyzed from two aspects including the scatter of fatigue data and the porosity in the additive manufacturing materials. The influence of the volume energy density ratio on the fatigue properties of additive manufacturing metal materials is discussed, and the results are analyzed. This research provides an effective method for the fatigue damage evaluation of additive manufacturing metal materials.

Target trajectory association method based on detection association and deep learning is one of the research hotspots in the field of computer vision. However, due to the lack of effective space-time constraints in the design of existing methods, and the insufficency of generalization ability of target apparent features, recognition errors will occur in the case of obvious differences in target orientation, which will lead to frequent ID switching and error association. To solve this problem, we propose a target trajectory association method based on orientation constraint and re-identification feature. This paper introduces pedestrian orientation discrimination into pedestrian re-identification, and presents a pedestrian re-identification network model with orientation constraint, which improves the representation ability of target features. Combining the spatial and temporal characteristics of target orientation, position information from Kalman filter and overlap area, a hierarchical trajectory association model based on orientation constraint is proposed to obtain the target trajectory in a single camera. A simple and effective bi-directional competitive matching mechanism is introduced to implement effective association of target trajectories in the cross camera scene. Experimental results show that the proposed method achieves a competitive level on MOT datasets. It can reduce frequent ID exchange, and can effectively solve the problem of error association when similar objects are moving towards each other. Meanwhile, with a frame rate of 19.6 frame/s, it can satisfy the requirements of near real-time scene.

In numerical simulation of foam microstructure, the geometric characteristics and arrangement of foam cavities have an important influence on calculation efficiency and calculation results. We propose a new algorithm, based on the advancing surface search geometric construction algorithm and the Laguerre partition algorithm, to generate the geometric model of the PVC foam microscopic specimen. First, reconstruct the authentic geometric model of the foam from μCT scan image, and measure the geometric characteristics of the foam cavity and the volume distribution pattern. Then, convert the measured foam cavity volume into a sphere, and put it into the space through the advancing surface search geometric construction algorithm. Finally, divide the space sphere into regions by means of Laguerre division, and assign wall thickness parameters to form a geometric model of the microstructure. The established model is in good agreement with the actual material in terms of micro-geometric characteristics.

In order to further improve the bearing capacity of plastic gears in the transmission of metal worm and plastic helical gears, the force characteristics of the traditional equal-pitch worm helical gear meshing transmission were studied in depth, and the unequal pitch meshing of the worm and the helical gear was innovatively proposed. Based on beam bending theory and gear tooth deformation theory, the relationship between tooth surface load, deformation and contact stiffness is obtained.On the premise that the bending deformation rate of the gear tooth root is equal, the theoretical design method of unequal meshing is deduced, and the pitch adjustment of the worm in the case of unequal pitch meshing is obtained, which is verified by static strength experiments.The experimental results show that the unequal pitch design can increase the load-bearing capacity of the plastic helical gear by 13.69%.

Automatic modulation recognition is a key technology for spatial cognitive communication system, which helps to realize adaptive signal demodulation. Although the deep neural network has the advantage of strong feature extraction, it suffers from the problems of numerous parameters and large amount of calculation, and thus is difficult to be implemented in in-orbit applications. To mitigate these problems, we propose a lightweight, high-performance convolutional neural network structure. The network first extracts the in-phase and quadrature features of the signal, then the time domain features, and finally the mean value of each channel feature for classification. The experimental results of the classification of 11 modulation methods show that when the signal-to-noise ratio is higher than 0 dB, the average recognition accuracy can reach 86.94%, which is 31.54% higher than that of traditional cumulant methods. Compared with the current deep neural network model with high recognition accuracy, the network proposed uses only less than 10% of model parameters, and increases the calculation speed by an average of 20 times on Raspberry Pi 4B.

Aiming at the problem of model data fusion among different design tools in complex product design, the research explores the fusion strategy of end-to-end heterogeneous model data between tools. A multi-layer collaborative strategy of heterogeneous data is proposed, which uses the dynamic characteristics of database management and model attribute sharing to realize the integration of heterogeneous model data. In the system integration environment of OpenMBEE, through the secondary development of the modeling tool CREO, the strategy is employed to obtain the dynamic model attribute information in the whole life cycle design. The effectiveness of the strategy is verified by 3D model editing and reuse function testing. In order to realize the fusion of heterogeneous data, an intelligent algorithm to automatically obtain the attribute information of visual model is explored, based on Transformer model and bi-directional LSTM (Bi-LSTM) model. Utilizing the multi-layer perceptual characteristics of neural network, the algorithm realizes automatic extraction of heterogeneous data attribute information through deep learning and feature analysis of the attribute information in the model. The effectiveness of the intelligent model information extraction is verified by the model data set that is established with the requirement analysis models designed by modeling tool CAMEO.

In the scientific research field, storage capacity, processing efficiency and analysis accuracy cannot keep pace with the exponential growth rate of scientific data. Thus, a massive scientific data calculation framework named BSDF is proposed based on scientific data structure and standards. A unified data interface based on model-driving is integrated to implement indiscriminate access to heterogeneous scientific data. Then an auto-partitioning algorithm based on scientific metadata is proposed, which determines task granularities through parameter prefetching and hyperplane dimension calculation. Experimental results show that compared with the performance of the H5Spark framework, that of the BSDF is increased by 39%-68% in nine benchmark tests. In the optimization of the domain-specific PKTM algorithm, a speedup ratio is increased by 41.62 times.

This paper proposes an autonomous dynamic obstacle avoidance method for omnidirectional mobile robot by introducing velocity repulsion field to solve the existing problems, an improvement from water flow field based artificial potential field obstacle avoidance method. This paper presents a detailed analysis of problems of artificial potential field improved by water flow field, such as a too long avoidance path or avoidance failure caused by the mobile robot moving in front of the obstacles. To solve the above problems, the velocity repulsion field, in line with the relative velocity of the mobile robot and the dynamic obstacle, is introduced in artificial potential field obstacle avoidance method based on water flow field. With the omni-directional mobile robot moving in rear of the obstacles, a safe and autonomous dynamic obstacle avoidance is fully realized. The effectiveness and practicability of the autonomous dynamic obstacle avoidance algorithm are verified through simulation and indoor obstacle avoidance experiment.

Fault diagnosis is an important technology to improve the reliability of the switched reluctance motor (SRM) speed control system. To address the non-linear and unstable fault signal of the switched reluctance motor power converter, and the problem that effective information is easily covered by noise, a new fault feature extraction method is proposed. The DC bus current is subjected to variational mode decomposition to obtain several intrinsic mode functions. The average value of the permutation entropy of the multi-scale effective modal components is taken as the feature vector, and is input into the support vector machine classifier for fault identification. In order to verify the feasibility of the proposed algorithm, a simulation model was established and compared with traditional fault diagnosis algorithms such as wavelet analysis; meanwhile, a switched reluctance motor experiment bench was built to test the open circuit and short circuit fault states. The simulation and experimental results show that the method proposed in this paper can reduce the influence of noise and improve the accuracy of fault identification rate.

Eye-tracking interaction is the key control method for head-mounted virtual reality (VR)/augmented reality (AR) devices and non-calibrated gaze estimation is one of the core problem in current VR/AR eye-tracking interactions. Efficient and robust non-calibrated gaze estimation requires a large amount of training data and an efficient network structure. Based on the existing deep-learning-based near-eye gaze estimation method, by adding multitasking auxiliary reasoning and increasing the multi-stage output of the network structure for joint multi-task training, we achieve an effective improvement of gaze estimation accuracy without increasing the refer time compared to the original gaze estimation network. During model training, multiple intermediate stages of the gaze estimation network structure are used to derive multiple parallel network headers for auxiliary reasoning about eye features, including semantic segmentation of eye images, iris border frames, and eye contour information, to provide multi-stage relay monitoring for the original gaze estimation network, which effectively improves the generalization capability of the gaze estimation network without increasing the training data. Experiments on the open datasets Acomo-14 and OpenEDS2020 show that the accuracy of the algorithm is improved by 21.74% and 18.91%, respectively, and the average gaze estimation error is reduced to 1.38 degrees and 2.01 degrees, compared with the network without auxiliary reasoning.

Aiming at the problem of multi-UAV cooperative path following in three-dimensional space, a cooperative controller was designed based on backstepping. To avoid control input saturation, an auxiliary controller was introduced to ensure control performance of the system. The six degrees of freedom nonlinear model of the UAV was feedback linearized, and the uncertain disturbances of the UAV and the unmodeled dynamics of the model itself were estimated and compensated online by radial basis function, thus improving the robustness of the system and its capacity to resist disturbances. Meanwhile, a first-order filter was introduced to avoid the derivation of the virtural control variable in the backstepping method. The graph theory was adopted to solve the communication problem between UAVs, and the consensus theory to realize the cooperative control of multiple UAVs. Finally, the Lyapunov stability theory was used to prove the stability of the system. The simulation results show that the designed cooperative path following controller can achieve good cooperative tracking control effects.

To reduce the drift of hemispherical resonator gyroscope (HRG) caused by temperature effect, a temperature drift compensation model is established, which compensates the deterministic drift related to temperature. An improved PSO-ARMA modeling method is proposed to compensate the uncertain drift. This method introduces the decreasing inertia weight strategy into the opposition-based learning particle swarm optimization (PSO) algorithm, improving the algorithm's ability to jump out of the local and converge fast. In the modeling process, the improved PSO algorithm is used to optimize the ARMA parameters, thus improving the model accuracy. The experimental data of HRG temperature rise are used for verification. After the model compensation, the output accuracy of HRG can reach 0.07°/h, which is twice as high as that of traditional ARMA modeling method.

To study the hydrodynamic lubrication reliability of journal bearings inside aero-gear pumps running at high speed and with low medium viscosity, an elastohydrodynamic lubrication (EHL) model considering elastic deformation of bearing bush is established by coupling Reynolds lubrication equation with influence matrix. Considering the uncertainty caused by the size tolerance and operating conditions of journal bearings, the peak pressure of hydrodynamic lubrication, which is closely related to the lubrication characteristics of journal bearings, is taken as the reliability criterion. The AK-IS method combining adaptive Kriging and importance sampling is used to calculate the reliability and sensitivity of the hydrodynamic lubrication characteristics of journal bearings. The results show that considering bearing deformation, the pressure peak value of journal bearings is 15.04% lower than that of rigid bearings, and hence the influence of EHL on hydrodynamic lubrication of bearings cannot be neglected. The results also show the accuracy and efficiency of the AK-IS based reliability analysis of hydrodynamic lubrication of journal bearings. Moreover, the degree of the effect of various uncertainties on the reliability of hydrodynamic lubrication is different, with the radius clearance of bearings being the most sensitive to reliability, and the rotational speed the least.

This paper studies a blended-wing-body aircraft with distributed propulsion, explores the influence of suction control (with suction position and suction momentum variables) on the aerodynamic characteristics of aircraft in take-off and cruise state and explains the mechanism of the influence of suction flow control on the aerodynamic characteristics of the blended-wing-body aircraft. The results show that under the condition of high take-off attack angle, compared with the non-aspirated state, the maximum lift coefficient of the aircraft is increased by 7.16% when the aspirator is located in the outer wing segment (chord position being 0.05c, and inspiratory momentum being 0.02). In the cruise state, when the aspirator is located in the centrosome (chord position being 0.6c, and inspiratory momentum being 0.012 5), the pressure distribution of the powertrain is improved, and the lift-drag ratio of the aircraft is increased by 2.14% compared with that of the non-getter state.

To overcome the disadvantages of the existing inertial stabilization platform such as large interference for using mechanical bearings, high difficulty for using air/liquid bearings and poor linearity for using magnetic resistance magnetic bearings, a new Lorentz inertial stability platform (LISP) based on Lorentz force deflection magnetic bearing is proposed. To suppress the influence of coupling effect and load-bearing friction resonance interference on the high-frequency attitude compensation control of the platform deflection channel, a digital control scheme based on LESO-PID combined with Kalman filter (KF) feedback is proposed. According to the structural characteristics of rotor tilt supported by Lorentz force magnetic bearing (LFMB), a dynamics model for LISP deflection is established; the tilting relationship of two radial channels is analyzed with the established model, and the linear extended state observer (LESO) and Kalman filter feedback control is introduced into PID controller to suppress friction resonance interference and coupling effects; a digital control system based on DSP and FPGA is construed, and the control method is digitalized in a discrete form. The stability of the proposed control method is analyzed by logarithmic frequency characteristic criterion and Nichols curve, and the stabilities of the rotor deflection channel before and after importing LESO-Kalman are compared through simulation. Experimental results show that with traditional PID, the rotor system causes serious distortion at high frequency, while the system reduces noise and interference greatly after importing LESO-Kalman control. Meanwhile, the internal state parameters of the system can be monitored in real time. Experimental results verify the effectiveness of proposed control method to suppress the frictional resonance interference and coupling effects.

The backstepping sliding mode control method based on extended state observer (ESO) is proposed, which can solve the compound disturbance problem caused by unmodeled friction force, parameter uncertainty and external random disturbance. The ESO is designed to estimate the velocity and acceleration of actuator. The backstepping sliding mode controller is designed based on the displacement feedback signal and the estimated values of ESO. By constructing Lyapunov function including backstepping design error, sliding mode function and observer error, the stability of the proposed control method is proved. In order to verify the effectiveness of the proposed method, AMESim and MATLAB/Simulink co-simulation is carried out to compare with PID controller, traditional backstepping sliding mode controller and sliding mode controller based on ESO, and the simulation data is analyzed. The results show that the proposed method can effectively suppress the compound disturbance of the system, with higher precision and stronger robustness.

For the under-actuated quadrotor UAV swarm systems with multiple leaders and followers, a distributed hierarchical formation-containment control method is proposed. First, a hierarchical distributed finite-time sliding mode estimator is designed to achieve that each UAV can generate estimated position information that meets the control needs under the condition that only some leaders can obtain the desired trajectory. Then, considering the research object is an under-actuated six-degree-of-freedom quadrotor UAV model, a hierarchical control method of the UAV position layer and the attitude layer is proposed, which realizes the tracking control of the generated estimated position. This method adopts a high-order derivative approximation algorithm to prevent differential explosions in the process of solving the desired angular velocity. The given method can realize the effective formation-containment under the condition of satisfying the stable convergence of attitude. Finally, the accuracy and effectiveness of the proposed method are verified through numerical simulation.

Aiming at the problem of slow convergence speed when calculating Lagrange multiplier by bisection method, the proposed Lagrange multiplier calculation method is applied to the optimal criteria (OC) method and the guide-weight (GW) method to update the density, and the results of the proposed method are compared with the bisection method. The topology optimization model with the smallest compliance under the volume constraint is established. The elastic modulus of element density is calculated by the solid isotropic material with penalization (SIMP) or rational approximation of material properties (RAMP) method. The multiplier is calculated by the proposed method in this paper, and the element density is updated by the GW method. The number of gray element is reduced by the Heaviside projection function. The computational results show that: the proposed method does not significantly reduce the number of finite element analysis, but the CPU time used by the proposed method to calculate the Lagrange multiplier is less than that of the bisection method, and the number of density updates is reduced to less than 50% than before. In addition, in the two numerical examples, when SIMP model is adopted, the structure obtained by the GW method has smaller compliance than the OC method.