Search

In large-scale complex system, The establishment of a fast, accurate fault diagnosis system is more difficult because there exist many uncertain elements between the fault cause and the fault sign .A fault diagnosis system is established based on RBF cloud neural network ,the RBR (rule-based reasoning) and the CBR (case-based reasoning).The fault diagnosis system not only has the advantages of self-learning, high accuracy, randomness, fuzziness, etc ,and has the advantages of independently of mathematical model ,rich knowledge representation, mighty problem solving ability, etc. Theoretical analysis and simulation results show that the system is feasible and effective for fast and accurate fault positioning of complex systems.

This paper provides a review of 3D aircraft object recognition methods based on 2D images. First, essentialities and advantages on 3D aircraft image recognition are analyzed. Second, in view of the aircraft image recognition system model, the traits of aircraft images are elaborated, then the methods of aircraft image recognition at every stage are discussed, especial the methods of feature extraction about aircraft image and the feasibilities, which some new methods about 3D object recognition are used to aircraft image recognition, are focused. At last, some issues about aircraft image recognition which should be further studied in the future are proposed.

We use Fourier descriptors (FD's) to represent shape in model-based pose estimation. Specific invariance and normalization requirements for shape descriptors concerning to pose estimation are proposed. FD's are improved to meet such requirements. Common issues and techniques for such application are focused on. Starting point of the shape contour is fixed to the upper left corner point. Distance pairs instead of edge coordinate pairs recast the shape sequence. Moving average filtering is proceeded to remove the noises of the shape sequence. Shape sequence is re-sampled to make it definite length. FD's amplitude is normalized to the range of 0 to 1. Variance of sequences between observed and library FD's is defined as the shape matching objective function. For simulation, we use a missile model of Milkshape3d format. Results show that by adopting the improved FD's, we can arrive at a pose estimation practically by randomly optimal search of library FD's.

Methods for pose estimation of flying objects are introduced. Among them is the model-based optical method. We focus on the feature description aspect in model-based method. Feature descriptors of chain codes, moments, Fourier descriptors are used for 2D silhouette or region description. Common issues and techniques, particularly representation and normalization, of such three kinds of descriptors in the application of model-based pose estimation are analyzed. We build a Matlab pose estimation framework to compare pose estimation procedures using different feature descriptors. A missile model of MilkShape 3D file format is created as the simulation object. Experiments concerning with the abilities of descriptors are proceeded to show the difference of these descriptors.

The accuracy of the Shack-Hartmann wavefront sensor (SHWS) for measuring the distortion wavefront is mainly dependent upon the measurement accuracy of the centroid of the focal spot. Many methods have been presented to improve the accuracy of the wavefront centroid measurement by weakening the influence of various noises, such as the photo noise, the read-out noise, the background noise, unevenness and instability of the light source, etc. In general, these methods mainly use the first moment centroid algorithm to calculate the centroid in the whole sub-aperture. In this paper, we present an improved centroid measurement approach that calculate the centroid of the focal spot more precisely using the higher moment centroid method in an optimized detection window. Based on the improved method, the effects of various noises out of the optimized detection window are almost eliminated; furthermore, the noise influences in the optimized detection window are also weakened due to the more contributions of the focal spot intensity. The experimental results demonstrate that the precision and repeatability of focal spot centroid are more prominent than the results obtained via other commonly centroid methods.

A quick and high-precision image center locating algorithm based on the difference between the corresponding ellipse's long and short axe method was proposed, to meeting the requirements for high precision and time-saving in real time beam automatic alignment process of the high power laser device. The maximum value of the spot's area is obtained by acting the Big Law Method, 3×3 Neighborhood Method and Eight-Neighborhood Method on the spot image in turn; Lengths of the long and the short axis are obtained by the analyzing the center distance of the maximum spot, and adjust the spot shape by the deviation of the ellipse's long and short axe until to the deviation approach to zero; Through analysis the deviation between the actual spot center and the corresponding reference position, the adjusting steps of stepping motor controlled by the closed-loop control method is obtained, and the automatic alignment is realized. This algorithm is verified in the high-power laser device, the experiment results shows that the alignment time of the main amplification of the optical path is shortened to 15 min, the near-field alignment accuracy is better than 0.2%, and the far-field pointing accuracy is better than 1 μrad, all of them meet the alignment requirements of the high-power laser device. ©, 2015, Chinese Optical Society. All right reserved.