There are a lot of joint structures on the surface of aircraft skin, and their quality has an important influence on the performance, life and reliability of aircraft. At present, there exist some problems such as large amount of data, low efficiency of data processing and difficulty at the data processing stage of the point cloud data of skin seam structure. A point cloud segmentation method of aircraft skin seam structure driven by measurement position points is proposed to improve the calculation efficiency of seam parameters. The seam edge curve features of the skin theoretical model are extracted and discretized into measured position points, which guide the construction of the space bounding box and drive the local segmentation of the point cloud data. The experimental results show that this method can achieve good segmentation effect for the point cloud of the seam structure.

The extent of digitalization is increasing in aeronautical manufacturing, but workers are still the main part engaged in production activities of assembly. Due to the complex structure, numerous parts and large workload, higher technical requirement for workers of assembly is placed. Projection indication technology is one of the important approaches to improve the degree of digitalization and the production efficiency of manual assembly process. Firstly, starting from the implementation principle, the applicability of projection indication in the field of assisted assembly is described. Secondly, three main technical contents including system calibration, visual sensing and projection mapping are explained. Then, the applications of projection indication in the field of aircraft manufacturing are presented and its development trends are outlined. Finally, technical challenges are summarized on the basis of sorting out the domestic research status.

In order to solve the problems of unintelligent detection of weld points, low detection accuracy and poor detection robustness of the robot system in the automated welding process of the internal anti-surge plate of aluminum alloy tanks of large pressure vessels, a multi-feature fusion 3D weld point real-time detection method for structural light visionguided robots is designed. First, the visual features are extracted from the workpiece to derive multiple regions of interest, then the multiple features are fused to obtain the 2D key points, and finally the 3D pre-weld points are determined by fast and unconstrained system calibration. Industrial field experiments show that the maximum error of 3D welding point extraction in camera coordinate system is 0.196 mm, the average error is 0.099 mm, and the average detection time is 0.09 s. The welding point detection is accurate, fast and intelligent, which can meet the industrial robot welding path planning and automatic welding tasks.

The throat area of the geometry-variable nozzle directly affects the performance of the aero-engine. The inner diameter micrometer is traditionally used to measure the throat size, which has some problems, such as low efficiency and droop of the nozzle seal affecting the measurement accuracy. Therefore, we put forward a calibration method for throat area of tunable circular nozzle based on laser interference. Firstly, a coaxial length measurement structure is designed which features built-in laser interference optical path. A pressure sensor is utilized to give real-time feedback to automatically control the thrust. Then the polygon throat area of the nozzle is calculated from the dimensional measurement data. Finally, the correlation model between throat area and engine control signal is established based on the least square method. The experimental result shows that the size measurement accuracy of the developed system is 41 μm, the area calibration accuracy is 0.98147, and the method features good simplicity and convenience.

The digitization of X-ray film and automatic detection of weld defects are of great significance for improving the production and processing quality and detection efficiency of large aerospace parts. In some specific scenes, the digital receiver cannot be used for X-ray detection, and the transformation of X-ray film into digital image is the premise of defect recognition. However, it is difficult to realize the high fidelity digitization of X-ray film by existing methods. In this paper, an algorithm of X-ray film defect detection based on depth learning is proposed. Firstly, based on full convolution neural network, the digital image with best exposure time on the X-ray film is automatically selected from the images with different exposure time. Then a defect detection network based on lightweight MoGaA network is designed to detect small target defects in X-ray digital images. The digitization and detection results show that the accuracy of this algorithm for weld defect detection can reach 96%, and good detection effect is obtained.

Aiming at the problems of low assembly accuracy and large assembly interference force of the deck and main frame during the robot satellite assembly stage, an online measurement and compensation method of satellite assembly error combining vision and force perception is proposed. An online measurement system for the assembly error of the satellite deck and the main frame is established by using the visual inspection device. The binocular calibration, robot hand-eye calibration, and relative pose calibration of other components have been completed. A method for compensating the assembly error of the satellite deck and main frame is proposed. The real-time measurement and accurate compensation of assembly error are realized. The robot terminal load identification and gravity compensation are completed by force sensing device. The interference force between the satellite deck and the main frame assembly is measured in real-time, and the satellite flexible assembly is realized. The experimental results show that the assembly error of the deck and main frame is controlled within 0.2 mm and the assembly interference force is less than 50 N after the online measurement and compensation of the satellite assembly error with the fusion of vision and force perception is adopted. It meets the precision requirements of satellite assembly, and proves the effectiveness and stability of the proposed method.

To meet high precision digital measuring of robot in-situ milling for aircraft assembly, the bilateral filtering processing and fast data simplification algorithms were studied for 3D point cloud software used to data acquisition and processing was developed, based on which 3D point cloud were processed including to denoise and simplify. The method of point cloud processing in this paper reduced the damage of filtering to boundary features, and saved effectively key features of complex components including round corner, rib and boundary of spatial curve. Taking typical samples as test objects, the methods of point cloud processing were tested to compare with CMM measurement results. Developed software realizes efficient integration with automation equipment, has high openness of software interface, avoids a lot of human-computer interaction, reduces data processing time, and meet the timeliness requirements of robot in-situ milling equipment.

Aiming at the heavy measurement task in the process of fighter manufacture and assemble, as well as the problems of heavy equipment and low efficiency in the existing methods of boresight for fighter plane weapon, a new method of boresight for fighter plane weapon based on the measurement principle of binocular vision is proposed. A photoelectric target composed of infrared LEDs is designed, which is connected with calibration station of the fighter through the adapter. The binocular cameras take pictures of all photoelectric targets, and then obtains the three-dimensional coordinates of the LED points. Then, the relative attitude parameters between the binocular cameras and each photoelectric target will be calculated by using the principle of space coordinate transformation based on least square method. And the horizontal/vertical deviation angles of the stations axes to be measured relative to the axis of the fighter’s reference station can be finally obtained by combining the coordinate system transfer, and the boresight operation will be completed. Based on the principle of boresight, a 3D simulation environment is built, the possible error sources in the process of boresight are analyzed, and the boresight algorithm is simulated using Monte Carlo method. The simulation results show that the  cameras calibration error has the greatest impact on the measurement accuracy of horizontal/vertical deviation angles, with the standard deviation of 0.0397° and 0.0268°, respectively. This proposed method can be applied to the boresight for fighter plane weapons, which can also improve the work efficiency and reliability in the process of assembly and measurement.

Light field imaging can simultaneously record the intensity and direction information of rays and has the ability of depth estimation. However, the current methods only achieve disparity estimation, and few methods convert the disparity to absolute depth to achieve three-dimensional measurement. Therefore, this paper theoretically analyzes the threedimensional imaging principle of lenslet-based light field cameras and establishes the numerical conversion relationship between disparity and absolute depth of lenslet-based light field cameras. Firstly, the relationship between the disparity and equivalent baseline is obtained by the theoretical analysis of the disparity principle in light field imaging. Secondly, the imaging principle of lenslet-based light field cameras is analyzed to obtain the relationship between the equivalent baseline and the camera’s internal parameters, and the numerical conversion relationship between disparity and absolute depth is obtained by combining the disparity and Gaussian imaging formula. Finally, the validity of the conversion relationship is verified by calibrating the light field camera and building an experimental platform. Experimental results show that this relationship can convert disparity into absolute depth and obtain high-precision results.

Underwater processing and manufacturing plays an important role in aviation, shipbuilding and other fields. In situ online 3-D detection of manufacturing process has become an urgent demand for underwater manufacturing quality assurance. Fringe projection profilometry (FPP), as one of the classic optical 3-D measurement technologies, holds the advantages of non-contact, high speed and high accuracy. However, in turbid water, due to the absorption and scattering of light, the fringe light intensity captured by the camera is attenuated, the contrast is reduced, the image details are blurred, and a lot of noise is introduced, resulting in poor fringe image quality. The phase calculated by the low-quality fringes has the non-negligible phase error, resulting in the decrease of the 3-D measurement accuracy. In order to reduce the influence of underwater absorption and scattering, an end-to-end fringe image enhancement algorithm based on deep learning is proposed. The fringe pattern enhancement convolutional neural network (FPENet) is used to convert the low contrast and high noise fringes into high contrast and low noise fringes to obtain more accurate phase results. FPENet can effectively improve fringe quality and reduce phase error for water with different turbidity. Especially in high turbidity water, the phase error can be reduced by about 50%, significantly improving the measurement accuracy of underwater FPP, which is of great significance for improving the applicability of FPP in complex scenes.

In the electrochemical machining (ECM) of the blisk, the flow field plays an important role in the machining stability, and the inlet angle of electrolyte solution has an important influence on the uniformity of the flow field. For the flow pattern from blade tip to blade root, five inlet angle flow models (12°, 22°, 32°, 42° and 52°) were designed, and the flow field simulation of electrochemical machining was carried out. The results show that the average flow rate is 19.01 m/s and the mean square deviation of flow rate is 6.33 when the inlet angle of electrolyte is 32°, which meets the requirements of flow field for electrochemical machining of the blisk. Under the flow field form with inlet angle of electrolyte solution of 32°, the electrochemical machining experiment of the blisk was carried out. The machining process was stable, the surface of the workpiece was free of flow texture, the machining accuracy was 0.12 mm, and the surface roughness was R_a0.353 μm, which verified the rationality of the flow field.

SiC_p/Al is a topical difficult-to-cut material. In the cutting progress, the exist of particulates reinforcements makes the machined surface of the material easily appear some defects like matrix tearing, microcrack and microvoid during cutting. In order to achieve high efficiency and low damage machining of SiC_p/Al composites, the paper summarizes macroscopic and microscopic modeling method which characterizes SiC_p/Al dynamic mechanical properties. It analyzes the hotspots and difficult problems of the multi-scale and multiphase coupling cutting finite element simulation. It summarizes the cutting performance of SiC_p/Al composite and its influencing factors from the perspectives of chip formation mechanism, surface integrity and tool wear. Research shows that removal mode and mechanism of the particulates reinforcements have a significant effect on surface formation. It discusses research progress of the mechanism of processing ductile regime and the acquiring critical conditions of brittle-ductile transition. It points out that low temperature micro cutting, ultrasonic assisted micro cutting and laser assisted micro cutting are the development directions which achieve compatible deformation and the processing conditions of ductile regime of SiC_p/Al composites.

In order to promote the application of laser cutting in the field of aviation, process research of the aviation aluminum alloy laser cutting has been carried out. Data samples of the process parameters and the quality characteristics were obtained by orthogonal test. An artificial neural network model was built by Python language to study the data samples, which has double hidden layer. Based on this model, the correlation between the process parameters and the quality characteristics were investigated and predicted. The result reveals that the predicted average error of the quality characteristics can be controlled in 10%. The size deviation, roughness and heat affected zone width of the test specimens optimized by the model were measured, and the errors between the actual measured value and the predicted value are 8.5%, 6.4% and 8.2% respectively.